Add support for new document page structure

Documents now have their own dedicated DocumentPage
submodel. The old DocumentPage is now called DocumentVersionPage.
This allows mappings between document pages and document version
pages, allowing renumbering, appending pages.
DocumentPages have a content_object to map them to any other
object. For now they only map to DocumentVersionPages.
New option added to the version upload form to append the
pages of the new version.
A new view was added to just append new pages with wraps the
new document version upload form and hides the append pages
checkbox set to True.
Add a new action, reset_pages to reset the pages of the
document to those of the latest version.

Missing: appending tests, checks for proper content_object in OCR and
document parsing.

Author: Roberto Rosario <roberto.rosario@mayan-edms.com>
Date:   Thu Oct 11 12:00:25 2019 -0400

docs/releases/3.3.rst

@@ -93,7 +93,7 @@ Changes
 Removals
 --------
 
-- Database conversion. Reason for removal. The database conversions support
+- Database conversion. Reason for removal: The database conversions support
   provided by this feature (SQLite to PostgreSQL) was being confused with
   database migrations and upgrades.
 
@@ -108,7 +108,7 @@ Removals
 
   Continued confusion about the purpose of the feature and confusion about
   how errors with this feature were a reflexion of the code quality of
-  Mayannecessitated the removal of the database conversion feature.
+  Mayan necessitated the removal of the database conversion feature.
 
 - Django environ
 

mayan/__init__.py

@@ -2,8 +2,8 @@ from __future__ import unicode_literals
 
 __title__ = 'Mayan EDMS'
 __version__ = '3.3beta1'
-__build__ = 0x030208
+__build__ = 0x030300
-__build_string__ = 'v3.2.8-255-g69086d87dd_Tue Oct 8 09:43:10 2019 -0400'
+__build_string__ = 'v3.3beta1_Sat Oct 5 15:08:53 2019 -0400'
 __django_version__ = '1.11'
 __author__ = 'Roberto Rosario'
 __author_email__ = 'roberto.rosario@mayan-edms.com'

mayan/apps/cabinets/apps.py

@@ -10,12 +10,14 @@ from mayan.apps.common.menus import (
     menu_facet, menu_list_facet, menu_main, menu_multi_item, menu_object,
     menu_secondary
 )
+from mayan.apps.documents.search import (
+    document_page_search, document_search, document_version_page_search
+)
 from mayan.apps.events.classes import ModelEventType
 from mayan.apps.events.links import (
     link_events_for_object, link_object_event_types_user_subcriptions_list,
 )
 from mayan.apps.events.permissions import permission_events_view
-from mayan.apps.documents.search import document_page_search, document_search
 from mayan.apps.navigation.classes import SourceColumn
 
 from .dependencies import *  # NOQA
@@ -115,12 +117,16 @@ class CabinetsApp(MayanAppConfig):
         )
 
         document_page_search.add_model_field(
-            field='document_version__document__cabinets__label',
+            field='document__cabinets__label',
             label=_('Cabinets')
         )
         document_search.add_model_field(
             field='cabinets__label', label=_('Cabinets')
         )
+        document_version_page_search.add_model_field(
+            field='document_version__document__cabinets__label',
+            label=_('Cabinets')
+        )
 
         menu_facet.bind_links(
             links=(link_document_cabinet_list,), sources=(Document,)

mayan/apps/common/generics.py

@@ -53,8 +53,8 @@ class MultiFormView(DjangoFormView):
     template_name = 'appearance/generic_form.html'
 
     def _create_form(self, form_name, klass):
-        form_kwargs = self.get_form_kwargs(form_name)
+        form_kwargs = self.get_form_kwargs(form_name=form_name)
-        form_create_method = 'create_%s_form' % form_name
+        form_create_method = 'create_{}_form'.format(form_name)
         if hasattr(self, form_create_method):
             form = getattr(self, form_create_method)(**form_kwargs)
         else:
@@ -71,7 +71,7 @@ class MultiFormView(DjangoFormView):
 
     def forms_valid(self, forms):
         for form_name, form in forms.items():
-            form_valid_method = '%s_form_valid' % form_name
+            form_valid_method = '{}_form_valid'.format(form_name)
 
             if hasattr(self, form_valid_method):
                 return getattr(self, form_valid_method)(form)
@@ -98,8 +98,8 @@ class MultiFormView(DjangoFormView):
 
     def get_form_kwargs(self, form_name):
         kwargs = {}
-        kwargs.update({'initial': self.get_initial(form_name)})
+        kwargs.update({'initial': self.get_initial(form_name=form_name)})
-        kwargs.update({'prefix': self.get_prefix(form_name)})
+        kwargs.update({'prefix': self.get_prefix(form_name=form_name)})
 
         if self.request.method in ('POST', 'PUT'):
             kwargs.update({
@@ -124,7 +124,7 @@ class MultiFormView(DjangoFormView):
         )
 
     def get_initial(self, form_name):
-        initial_method = 'get_%s_initial' % form_name
+        initial_method = 'get_{}_initial'.format(form_name)
         if hasattr(self, initial_method):
             return getattr(self, initial_method)()
         else:

mayan/apps/document_comments/apps.py

@@ -8,7 +8,9 @@ from mayan.apps.common.apps import MayanAppConfig
 from mayan.apps.common.menus import (
     menu_facet, menu_list_facet, menu_object, menu_secondary
 )
-from mayan.apps.documents.search import document_page_search, document_search
+from mayan.apps.documents.search import (
+    document_page_search, document_search, document_version_page_search
+)
 from mayan.apps.events.classes import ModelEventType
 from mayan.apps.events.links import (
     link_events_for_object, link_object_event_types_user_subcriptions_list
@@ -80,13 +82,17 @@ class DocumentCommentsApp(MayanAppConfig):
         SourceColumn(attribute='comment', source=Comment)
 
         document_page_search.add_model_field(
-            field='document_version__document__comments__comment',
+            field='document__comments__comment',
             label=_('Comments')
         )
         document_search.add_model_field(
             field='comments__comment',
             label=_('Comments')
         )
+        document_version_page_search.add_model_field(
+            field='document_version__document__comments__comment',
+            label=_('Comments')
+        )
 
         menu_facet.bind_links(
             links=(link_comments_for_document,), sources=(Document,)

mayan/apps/document_parsing/admin.py

@@ -3,13 +3,13 @@ from __future__ import unicode_literals
 from django.contrib import admin
 
 from .models import (
-    DocumentPageContent, DocumentVersionParseError
+    DocumentVersionPageContent, DocumentVersionParseError
 )
 
 
-@admin.register(DocumentPageContent)
+@admin.register(DocumentVersionPageContent)
-class DocumentPageContentAdmin(admin.ModelAdmin):
+class DocumentVersionPageContentAdmin(admin.ModelAdmin):
-    list_display = ('document_page',)
+    list_display = ('document_version_page',)
 
 
 @admin.register(DocumentVersionParseError)

mayan/apps/document_parsing/api_views.py

@@ -8,7 +8,7 @@ from rest_framework.response import Response
 from mayan.apps.documents.models import Document
 from mayan.apps.rest_api.permissions import MayanPermission
 
-from .models import DocumentPageContent
+from .models import DocumentVersionPageContent
 from .permissions import permission_content_view
 from .serializers import DocumentPageContentSerializer
 
@@ -41,8 +41,8 @@ class APIDocumentPageContentView(generics.RetrieveAPIView):
 
         try:
             content = instance.content
-        except DocumentPageContent.DoesNotExist:
+        except DocumentVersionPageContent.DoesNotExist:
-            content = DocumentPageContent.objects.none()
+            content = DocumentVersionPageContent.objects.none()
 
         serializer = self.get_serializer(content)
         return Response(serializer.data)

mayan/apps/document_parsing/apps.py

@@ -12,7 +12,9 @@ from mayan.apps.common.classes import ModelField
 from mayan.apps.common.menus import (
     menu_facet, menu_list_facet, menu_multi_item, menu_secondary, menu_tools
 )
-from mayan.apps.documents.search import document_search, document_page_search
+from mayan.apps.documents.search import (
+    document_search, document_page_search, document_version_page_search
+)
 from mayan.apps.documents.signals import post_version_upload
 from mayan.apps.events.classes import ModelEventType
 from mayan.apps.navigation.classes import SourceColumn
@@ -43,7 +45,7 @@ from .permissions import (
     permission_parse_document
 )
 from .signals import post_document_version_parsing
-from .utils import get_document_content
+from .utils import get_document_content, get_document_version_content
 
 logger = logging.getLogger(__name__)
 
@@ -74,6 +76,9 @@ class DocumentParsingApp(MayanAppConfig):
         DocumentVersion = apps.get_model(
             app_label='documents', model_name='DocumentVersion'
         )
+        DocumentVersionPage = apps.get_model(
+            app_label='documents', model_name='DocumentVersionPage'
+        )
         DocumentVersionParseError = self.get_model(
             model_name='DocumentVersionParseError'
         )
@@ -85,7 +90,7 @@ class DocumentParsingApp(MayanAppConfig):
             name='content', value=get_document_content
         )
         DocumentVersion.add_to_class(
-            name='content', value=get_document_content
+            name='content', value=get_document_version_content
         )
         DocumentVersion.add_to_class(
             name='submit_for_parsing',
@@ -100,9 +105,9 @@ class DocumentParsingApp(MayanAppConfig):
             )
         )
 
-        ModelField(
+        #ModelField(
-            model=Document, name='versions__version_pages__content__content'
+        #    model=Document, name='versions__pages__content__content'
-        )
+        #)
 
         ModelPermission.register(
             model=Document, permissions=(
@@ -133,17 +138,17 @@ class DocumentParsingApp(MayanAppConfig):
         )
 
         document_search.add_model_field(
-            field='versions__version_pages__content__content', label=_('Content')
+            field='versions__pages__content__content', label=_('Content')
         )
 
-        document_page_search.add_model_field(
+        document_version_page_search.add_model_field(
             field='content__content', label=_('Content')
         )
 
         menu_facet.bind_links(
             links=(link_document_content,), sources=(Document,)
         )
-        menu_facet.bind_links(
+        menu_list_facet.bind_links(
             links=(link_document_page_content,), sources=(DocumentPage,)
         )
         menu_list_facet.bind_links(

mayan/apps/document_parsing/forms.py

@@ -8,7 +8,7 @@ from django.utils.translation import ugettext_lazy as _, ugettext
 
 from mayan.apps.common.widgets import TextAreaDiv
 
-from .models import DocumentPageContent
+from .models import DocumentVersionPageContent
 
 
 class DocumentContentForm(forms.Form):
@@ -26,10 +26,10 @@ class DocumentContentForm(forms.Form):
         except AttributeError:
             document_pages = []
 
-        for page in document_pages:
+        for document_page in document_pages:
             try:
-                page_content = page.content.content
+                page_content = document_page.content_object.content.content
-            except DocumentPageContent.DoesNotExist:
+            except DocumentVersionPageContent.DoesNotExist:
                 pass
             else:
                 content.append(conditional_escape(force_text(page_content)))
@@ -37,7 +37,7 @@ class DocumentContentForm(forms.Form):
                     '\n\n\n<hr/><div class="document-page-content-divider">- %s -</div><hr/>\n\n\n' % (
                         ugettext(
                             'Page %(page_number)d'
-                        ) % {'page_number': page.page_number}
+                        ) % {'page_number': document_page.page_number}
                     )
                 )
 
@@ -72,8 +72,8 @@ class DocumentPageContentForm(forms.Form):
         self.fields['contents'].initial = ''
 
         try:
-            page_content = document_page.content.content
+            page_content = document_page.content_object.content.content
-        except DocumentPageContent.DoesNotExist:
+        except DocumentVersionPageContent.DoesNotExist:
             pass
         else:
             content = conditional_escape(force_text(page_content))

mayan/apps/document_parsing/icons.py

@@ -17,6 +17,9 @@ icon_document_content_download = Icon(
 icon_document_multiple_submit = Icon(
     driver_name='fontawesome', symbol='font'
 )
+icon_document_page_content = Icon(
+    driver_name='fontawesome', symbol='font'
+)
 icon_document_submit = Icon(
     driver_name='fontawesome', symbol='font'
 )

mayan/apps/document_parsing/links.py

@@ -32,9 +32,15 @@ link_document_content_delete_multiple = Link(
     text=_('Delete parsed content'),
     view='document_parsing:document_content_delete_multiple',
 )
+link_document_content_download = Link(
+    args='resolved_object.id',
+    icon_class_path='mayan.apps.document_parsing.icons.icon_document_content_download',
+    permissions=(permission_content_view,), text=_('Download content'),
+    view='document_parsing:document_content_download'
+)
 link_document_page_content = Link(
     args='resolved_object.id', conditional_disable=is_document_page_disabled,
-    icon_class_path='mayan.apps.document_parsing.icons.icon_document_content',
+    icon_class_path='mayan.apps.document_parsing.icons.icon_document_page_content',
     permissions=(permission_content_view,), text=_('Content'),
     view='document_parsing:document_page_content'
 )
@@ -44,12 +50,6 @@ link_document_parsing_errors_list = Link(
     permissions=(permission_content_view,), text=_('Parsing errors'),
     view='document_parsing:document_parsing_error_list'
 )
-link_document_content_download = Link(
-    args='resolved_object.id',
-    icon_class_path='mayan.apps.document_parsing.icons.icon_document_content_download',
-    permissions=(permission_content_view,), text=_('Download content'),
-    view='document_parsing:document_content_download'
-)
 link_document_submit_multiple = Link(
     icon_class_path='mayan.apps.document_parsing.icons.icon_document_submit',
     text=_('Submit for parsing'),

mayan/apps/document_parsing/managers.py

@@ -18,11 +18,13 @@ from .signals import post_document_version_parsing
 logger = logging.getLogger(__name__)
 
 
-class DocumentPageContentManager(models.Manager):
+class DocumentVersionPageContentManager(models.Manager):
     def delete_content_for(self, document, user=None):
         with transaction.atomic():
             for document_page in document.pages.all():
-                self.filter(document_page=document_page).delete()
+                self.filter(
+                    document_version_page=document_page.content_object
+                ).delete()
 
             event_parsing_document_content_deleted.commit(
                 actor=user, target=document

mayan/apps/document_parsing/migrations/0005_rename_page_content.py

@@ -0,0 +1,42 @@
+from __future__ import unicode_literals
+
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ('document_parsing', '0004_auto_20180917_0645'),
+        ('documents', '0052_rename_document_page'),
+    ]
+
+    operations = [
+        migrations.RenameModel(
+            'DocumentPageContent', 'DocumentVersionPageContent'
+        ),
+        migrations.AlterField(
+            model_name='documentversionpagecontent',
+            name='document_page',
+            field=models.OneToOneField(
+                on_delete=django.db.models.deletion.CASCADE,
+                related_name='content',
+                to='documents.DocumentVersionPage',
+                verbose_name='Document version page'
+            ),
+        ),
+        migrations.RenameField(
+            model_name='documentversionpagecontent',
+            old_name='document_page',
+            new_name='document_version_page',
+        ),
+        migrations.AlterModelOptions(
+            name='documentversionpagecontent',
+            options={
+                'verbose_name': 'Document version page content',
+                'verbose_name_plural': 'Document version pages contents'
+            },
+        ),
+    ]
+
+
+

mayan/apps/document_parsing/models.py

@@ -5,36 +5,12 @@ from django.utils.encoding import force_text, python_2_unicode_compatible
 from django.utils.translation import ugettext_lazy as _
 
 from mayan.apps.documents.models import (
-    DocumentPage, DocumentType, DocumentVersion
+    DocumentPage, DocumentType, DocumentVersion, DocumentVersionPage
 )
 
-from .managers import DocumentPageContentManager, DocumentTypeSettingsManager
+from .managers import (
-
+    DocumentVersionPageContentManager, DocumentTypeSettingsManager
-
+)
-@python_2_unicode_compatible
-class DocumentPageContent(models.Model):
-    """
-    This model store's the parsed content of a document page.
-    """
-    document_page = models.OneToOneField(
-        on_delete=models.CASCADE, related_name='content', to=DocumentPage,
-        verbose_name=_('Document page')
-    )
-    content = models.TextField(
-        blank=True, help_text=_(
-            'The actual text content as extracted by the document '
-            'parsing backend.'
-        ), verbose_name=_('Content')
-    )
-
-    objects = DocumentPageContentManager()
-
-    class Meta:
-        verbose_name = _('Document page content')
-        verbose_name_plural = _('Document pages contents')
-
-    def __str__(self):
-        return force_text(self.document_page)
 
 
 class DocumentTypeSettings(models.Model):
@@ -62,6 +38,32 @@ class DocumentTypeSettings(models.Model):
         verbose_name_plural = _('Document types settings')
 
 
+@python_2_unicode_compatible
+class DocumentVersionPageContent(models.Model):
+    """
+    This model store's the parsed content of a document page.
+    """
+    document_version_page = models.OneToOneField(
+        on_delete=models.CASCADE, related_name='content',
+        to=DocumentVersionPage, verbose_name=_('Document version page')
+    )
+    content = models.TextField(
+        blank=True, help_text=_(
+            'The actual text content as extracted by the document '
+            'parsing backend.'
+        ), verbose_name=_('Content')
+    )
+
+    objects = DocumentVersionPageContentManager()
+
+    class Meta:
+        verbose_name = _('Document version page content')
+        verbose_name_plural = _('Document version pages contents')
+
+    def __str__(self):
+        return force_text(self.document_page)
+
+
 @python_2_unicode_compatible
 class DocumentVersionParseError(models.Model):
     """

mayan/apps/document_parsing/parsers.py

@@ -23,11 +23,13 @@ class Parser(object):
     _registry = {}
 
     @classmethod
-    def parse_document_page(cls, document_page):
+    def parse_document_version_page(cls, document_version_page):
-        for parser_class in cls._registry.get(document_page.document_version.mimetype, ()):
+        for parser_class in cls._registry.get(document_version_page.document_version.mimetype, ()):
             try:
                 parser = parser_class()
-                parser.process_document_page(document_page)
+                parser.process_document_page(
+                    document_version_page=document_version_page
+                )
             except ParserError:
                 # If parser raises error, try next parser in the list
                 pass
@@ -41,7 +43,9 @@ class Parser(object):
         for parser_class in cls._registry.get(document_version.mimetype, ()):
             try:
                 parser = parser_class()
-                parser.process_document_version(document_version)
+                parser.process_document_version(
+                    document_version=document_version
+                )
             except ParserError:
                 # If parser raises error, try next parser in the list
                 pass
@@ -64,29 +68,33 @@ class Parser(object):
         )
         logger.debug('document version: %d', document_version.pk)
 
-        for document_page in document_version.pages.all():
+        for document_version_page in document_version.pages.all():
-            self.process_document_page(document_page=document_page)
+            self.process_document_version_page(
+                document_version_page=document_version_page
+            )
 
-    def process_document_page(self, document_page):
+    def process_document_version_page(self, document_version_page):
-        DocumentPageContent = apps.get_model(
+        DocumentVersionPageContent = apps.get_model(
-            app_label='document_parsing', model_name='DocumentPageContent'
+            app_label='document_parsing',
+            model_name='DocumentVersionPageContent'
         )
 
         logger.info(
             'Processing page: %d of document version: %s',
-            document_page.page_number, document_page.document_version
+            document_version_page.page_number,
+            document_version_page.document_version
         )
 
-        file_object = document_page.document_version.get_intermediate_file()
+        file_object = document_version_page.document_version.get_intermediate_file()
 
         try:
-            document_page_content, created = DocumentPageContent.objects.get_or_create(
+            document_version_page_content, created = DocumentVersionPageContent.objects.get_or_create(
-                document_page=document_page
+                document_version_page=document_version_page
             )
-            document_page_content.content = self.execute(
+            document_version_page_content.content = self.execute(
-                file_object=file_object, page_number=document_page.page_number
+                file_object=file_object, page_number=document_version_page.page_number
             )
-            document_page_content.save()
+            document_version_page_content.save()
         except Exception as exception:
             error_message = _('Exception parsing page; %s') % exception
             logger.error(error_message)
@@ -96,7 +104,8 @@ class Parser(object):
 
         logger.info(
             'Finished processing page: %d of document version: %s',
-            document_page.page_number, document_page.document_version
+            document_version_page.page_number,
+            document_version_page.document_version
         )
 
     def execute(self, file_object, page_number):

mayan/apps/document_parsing/serializers.py

@@ -2,10 +2,10 @@ from __future__ import unicode_literals
 
 from rest_framework import serializers
 
-from .models import DocumentPageContent
+from .models import DocumentVersionPageContent
 
 
 class DocumentPageContentSerializer(serializers.ModelSerializer):
     class Meta:
         fields = ('content',)
-        model = DocumentPageContent
+        model = DocumentVersionPageContent

mayan/apps/document_parsing/tasks.py

@@ -14,8 +14,8 @@ def task_parse_document_version(document_version_pk):
     DocumentVersion = apps.get_model(
         app_label='documents', model_name='DocumentVersion'
     )
-    DocumentPageContent = apps.get_model(
+    DocumentVersionPageContent = apps.get_model(
-        app_label='document_parsing', model_name='DocumentPageContent'
+        app_label='document_parsing', model_name='DocumentVersionPageContent'
     )
 
     document_version = DocumentVersion.objects.get(
@@ -24,6 +24,6 @@ def task_parse_document_version(document_version_pk):
     logger.info(
         'Starting parsing for document version: %s', document_version
     )
-    DocumentPageContent.objects.process_document_version(
+    DocumentVersionPageContent.objects.process_document_version(
         document_version=document_version
     )

mayan/apps/document_parsing/tests/test_events.py

@@ -10,7 +10,7 @@ from ..events import (
     event_parsing_document_version_submit,
     event_parsing_document_version_finish
 )
-from ..models import DocumentPageContent
+from ..models import DocumentVersionPageContent
 
 
 class DocumentParsingEventsTestCase(GenericDocumentTestCase):
@@ -19,7 +19,7 @@ class DocumentParsingEventsTestCase(GenericDocumentTestCase):
 
     def test_document_content_deleted_event(self):
         Action.objects.all().delete()
-        DocumentPageContent.objects.delete_content_for(
+        DocumentVersionPageContent.objects.delete_content_for(
             document=self.test_document
         )
 

mayan/apps/document_parsing/tests/test_parsers.py

@@ -18,5 +18,5 @@ class ParserTestCase(DocumentTestMixin, BaseTestCase):
         parser.process_document_version(self.test_document.latest_version)
 
         self.assertTrue(
-            TEST_DOCUMENT_CONTENT in self.test_document.pages.first().content.content
+            TEST_DOCUMENT_CONTENT in self.test_document.pages.first().content_object.content.content
         )

mayan/apps/document_parsing/tests/test_views.py

@@ -5,7 +5,7 @@ from django.test import override_settings
 from mayan.apps.documents.tests.base import GenericDocumentViewTestCase
 from mayan.apps.documents.tests.literals import TEST_HYBRID_DOCUMENT
 
-from ..models import DocumentPageContent
+from ..models import DocumentVersionPageContent
 from ..permissions import (
     permission_content_view, permission_document_type_parsing_setup,
     permission_parse_document
@@ -72,8 +72,8 @@ class DocumentContentViewsTestCase(
         self.assertEqual(response.status_code, 404)
 
         self.assertTrue(
-            DocumentPageContent.objects.filter(
+            DocumentVersionPageContent.objects.filter(
-                document_page=self.test_document.pages.first()
+                document_version_page=self.test_document.pages.first().content_object
             ).exists()
         )
 
@@ -86,8 +86,8 @@ class DocumentContentViewsTestCase(
         self.assertEqual(response.status_code, 302)
 
         self.assertFalse(
-            DocumentPageContent.objects.filter(
+            DocumentVersionPageContent.objects.filter(
-                document_page=self.test_document.pages.first()
+                document_version_page=self.test_document.pages.first().content_object
             ).exists()
         )
 

mayan/apps/document_parsing/urls.py

@@ -7,7 +7,9 @@ from .views import (
     DocumentContentView, DocumentContentDeleteView,
     DocumentContentDownloadView, DocumentPageContentView,
     DocumentParsingErrorsListView, DocumentSubmitView,
-    DocumentTypeSettingsEditView, DocumentTypeSubmitView, ParseErrorListView
+    DocumentTypeSettingsEditView, DocumentTypeSubmitView,
+    DocumentVersionPageContentView,
+    ParseErrorListView
 )
 
 urlpatterns = [
@@ -34,6 +36,11 @@ urlpatterns = [
         regex=r'^documents/pages/(?P<pk>\d+)/content/$',
         view=DocumentPageContentView.as_view(), name='document_page_content'
     ),
+    url(
+        regex=r'^documents/versions/pages/(?P<pk>\d+)/content/$',
+        view=DocumentVersionPageContentView.as_view(),
+        name='document_version_page_content'
+    ),
     url(
         regex=r'^documents/(?P<pk>\d+)/submit/$',
         view=DocumentSubmitView.as_view(), name='document_submit'

mayan/apps/document_parsing/utils.py

@@ -6,14 +6,28 @@ from django.utils.html import conditional_escape
 
 
 def get_document_content(document):
-    DocumentPageContent = apps.get_model(
+    DocumentVersionPageContent = apps.get_model(
-        app_label='document_parsing', model_name='DocumentPageContent'
+        app_label='document_parsing', model_name='DocumentVersionPageContent'
     )
 
-    for page in document.pages.all():
+    for document_page in document.pages.all():
         try:
-            page_content = page.content.content
+            page_content = document_page.content_object.content.content
-        except DocumentPageContent.DoesNotExist:
+        except DocumentVersionPageContent.DoesNotExist:
+            pass
+        else:
+            yield conditional_escape(force_text(page_content))
+
+
+def get_document_version_content(document_version):
+    DocumentVersionPageContent = apps.get_model(
+        app_label='document_parsing', model_name='DocumentVersionPageContent'
+    )
+
+    for document_version_page in document_version.pages.all():
+        try:
+            page_content = document_version_page.content.content
+        except DocumentVersionPageContent.DoesNotExist:
             pass
         else:
             yield conditional_escape(force_text(page_content))

mayan/apps/document_parsing/views.py

@@ -12,10 +12,12 @@ from mayan.apps.common.generics import (
 )
 from mayan.apps.common.mixins import ExternalObjectMixin
 from mayan.apps.documents.forms import DocumentTypeFilteredSelectForm
-from mayan.apps.documents.models import Document, DocumentPage, DocumentType
+from mayan.apps.documents.models import (
+    Document, DocumentPage, DocumentType, DocumentVersionPage
+)
 
 from .forms import DocumentContentForm, DocumentPageContentForm
-from .models import DocumentPageContent, DocumentVersionParseError
+from .models import DocumentVersionPageContent, DocumentVersionParseError
 from .permissions import (
     permission_content_view, permission_document_type_parsing_setup,
     permission_parse_document
@@ -46,7 +48,7 @@ class DocumentContentDeleteView(MultipleObjectConfirmActionView):
         return result
 
     def object_action(self, form, instance):
-        DocumentPageContent.objects.delete_content_for(
+        DocumentVersionPageContent.objects.delete_content_for(
             document=instance, user=self.request.user
         )
 
@@ -107,6 +109,30 @@ class DocumentPageContentView(SingleObjectDetailView):
         }
 
 
+class DocumentVersionPageContentView(SingleObjectDetailView):
+    form_class = DocumentPageContentForm
+    model = DocumentVersionPage
+    object_permission = permission_content_view
+
+    def dispatch(self, request, *args, **kwargs):
+        result = super(DocumentPageContentView, self).dispatch(
+            request, *args, **kwargs
+        )
+        self.get_object().document.add_as_recent_document_for_user(
+            request.user
+        )
+        return result
+
+    def get_extra_context(self):
+        return {
+            'hide_labels': True,
+            'object': self.get_object(),
+            'title': _(
+                'Content for document version page: %s'
+            ) % self.get_object(),
+        }
+
+
 class DocumentParsingErrorsListView(SingleObjectListView):
     view_permission = permission_content_view
 

mayan/apps/documents/admin.py

@@ -3,18 +3,12 @@ from __future__ import unicode_literals
 from django.contrib import admin
 
 from .models import (
-    DeletedDocument, Document, DocumentPage, DocumentType,
+    DeletedDocument, Document, DocumentType, DocumentTypeFilename,
-    DocumentTypeFilename, DocumentVersion, DuplicatedDocument, RecentDocument
+    DocumentVersion, DocumentVersionPage, DuplicatedDocument,
+    RecentDocument
 )
 
 
-class DocumentPageInline(admin.StackedInline):
-    model = DocumentPage
-    extra = 1
-    classes = ('collapse-open',)
-    allow_add = True
-
-
 class DocumentTypeFilenameInline(admin.StackedInline):
     model = DocumentTypeFilename
     extra = 1
@@ -29,6 +23,13 @@ class DocumentVersionInline(admin.StackedInline):
     allow_add = True
 
 
+class DocumentVersionPageInline(admin.StackedInline):
+    model = DocumentVersionPage
+    extra = 1
+    classes = ('collapse-open',)
+    allow_add = True
+
+
 @admin.register(DeletedDocument)
 class DeletedDocumentAdmin(admin.ModelAdmin):
     date_hierarchy = 'deleted_date_time'

mayan/apps/documents/api_views.py

@@ -33,10 +33,14 @@ from .serializers import (
     DocumentTypeSerializer, DocumentVersionSerializer,
     NewDocumentSerializer, NewDocumentVersionSerializer,
     RecentDocumentSerializer, WritableDocumentSerializer,
-    WritableDocumentTypeSerializer, WritableDocumentVersionSerializer
+    WritableDocumentTypeSerializer, WritableDocumentVersionSerializer,
+    DocumentVersionPageSerializer
 )
 from .settings import settings_document_page_image_cache_time
-from .tasks import task_generate_document_page_image
+from .tasks import (
+    task_generate_document_page_image,
+    task_generate_document_version_page_image
+)
 
 logger = logging.getLogger(__name__)
 
@@ -168,13 +172,8 @@ class APIDocumentPageImageView(generics.RetrieveAPIView):
         )
         return document
 
-    def get_document_version(self):
-        return get_object_or_404(
-            self.get_document().versions.all(), pk=self.kwargs['version_pk']
-        )
-
     def get_queryset(self):
-        return self.get_document_version().pages_all.all()
+        return self.get_document().pages_all.all()
 
     def get_serializer(self, *args, **kwargs):
         return None
@@ -221,6 +220,95 @@ class APIDocumentPageImageView(generics.RetrieveAPIView):
             return response
 
 
+class APIDocumentVersionPageImageView(generics.RetrieveAPIView):
+    """
+    get: Returns an image representation of the selected document version page.
+    """
+    lookup_url_kwarg = 'page_pk'
+
+    def get_document(self):
+        if self.request.method == 'GET':
+            permission_required = permission_document_view
+        else:
+            permission_required = permission_document_edit
+
+        document = get_object_or_404(Document.passthrough, pk=self.kwargs['pk'])
+
+        AccessControlList.objects.check_access(
+            obj=document, permissions=(permission_required,),
+            user=self.request.user
+        )
+        return document
+
+    def get_document_version(self):
+        return get_object_or_404(
+            self.get_document().versions.all(), pk=self.kwargs['version_pk']
+        )
+
+    def get_queryset(self):
+        return self.get_document_version().pages.all()
+
+    def get_serializer(self, *args, **kwargs):
+        return None
+
+    def get_serializer_class(self):
+        return None
+
+    @cache_control(private=True)
+    def retrieve(self, request, *args, **kwargs):
+        width = request.GET.get('width')
+        height = request.GET.get('height')
+        zoom = request.GET.get('zoom')
+
+        if zoom:
+            zoom = int(zoom)
+
+        rotation = request.GET.get('rotation')
+
+        if rotation:
+            rotation = int(rotation)
+
+        maximum_layer_order = request.GET.get('maximum_layer_order')
+        if maximum_layer_order:
+            maximum_layer_order = int(maximum_layer_order)
+
+        task = task_generate_document_version_page_image.apply_async(
+            kwargs=dict(
+                document_version_page_id=self.get_object().pk, width=width,
+                height=height, zoom=zoom, rotation=rotation,
+                maximum_layer_order=maximum_layer_order,
+                user_id=request.user.pk
+            )
+        )
+
+        cache_filename = task.get(timeout=DOCUMENT_IMAGE_TASK_TIMEOUT)
+        cache_file = self.get_object().cache_partition.get_file(filename=cache_filename)
+        with cache_file.open() as file_object:
+            response = HttpResponse(file_object.read(), content_type='image')
+            if '_hash' in request.GET:
+                patch_cache_control(
+                    response=response,
+                    max_age=settings_document_page_image_cache_time.value
+                )
+            return response
+
+
+class APIDocumentPageListView(generics.ListAPIView):
+    serializer_class = DocumentPageSerializer
+
+    def get_document(self):
+        document = get_object_or_404(Document, pk=self.kwargs['pk'])
+
+        AccessControlList.objects.check_access(
+            obj=document, permissions=(permission_document_view,),
+            user=self.request.user
+        )
+        return document
+
+    def get_queryset(self):
+        return self.get_document().pages.all()
+
+
 class APIDocumentPageView(generics.RetrieveUpdateAPIView):
     """
     get: Returns the selected document page details.
@@ -230,6 +318,33 @@ class APIDocumentPageView(generics.RetrieveUpdateAPIView):
     lookup_url_kwarg = 'page_pk'
     serializer_class = DocumentPageSerializer
 
+    def get_document(self):
+        if self.request.method == 'GET':
+            permission_required = permission_document_view
+        else:
+            permission_required = permission_document_edit
+
+        document = get_object_or_404(Document, pk=self.kwargs['pk'])
+
+        AccessControlList.objects.check_access(
+            obj=document, permissions=(permission_required,),
+            user=self.request.user
+        )
+        return document
+
+    def get_queryset(self):
+        return self.get_document().pages.all()
+
+
+class APIDocumentVersionPageView(generics.RetrieveUpdateAPIView):
+    """
+    get: Returns the selected document verion page details.
+    patch: Edit the selected document version page.
+    put: Edit the selected document version page.
+    """
+    lookup_url_kwarg = 'page_pk'
+    serializer_class = DocumentVersionPageSerializer
+
     def get_document(self):
         if self.request.method == 'GET':
             permission_required = permission_document_view
@@ -289,8 +404,7 @@ class APIDocumentTypeView(generics.RetrieveUpdateDestroyAPIView):
         'GET': (permission_document_type_view,),
         'PUT': (permission_document_type_edit,),
         'PATCH': (permission_document_type_edit,),
-        'DELETE': (permission_document_type_delete,)
+        'DELETE': (permission_document_type_delete,)    }
-    }
     permission_classes = (MayanPermission,)
     queryset = DocumentType.objects.all()
 
@@ -423,7 +537,7 @@ class APIRecentDocumentListView(generics.ListAPIView):
 
 
 class APIDocumentVersionPageListView(generics.ListAPIView):
-    serializer_class = DocumentPageSerializer
+    serializer_class = DocumentVersionPageSerializer
 
     def get_document(self):
         document = get_object_or_404(Document, pk=self.kwargs['pk'])

mayan/apps/documents/apps.py

@@ -59,7 +59,7 @@ from .links import (
     link_document_multiple_delete, link_document_multiple_document_type_edit,
     link_document_multiple_download, link_document_multiple_favorites_add,
     link_document_multiple_favorites_remove, link_document_multiple_restore,
-    link_document_multiple_trash, link_document_multiple_update_page_count,
+    link_document_multiple_trash, link_document_multiple_pages_reset,
     link_document_page_disable, link_document_page_multiple_disable,
     link_document_page_enable, link_document_page_multiple_enable,
     link_document_page_navigation_first, link_document_page_navigation_last,
@@ -74,8 +74,10 @@ from .links import (
     link_document_type_filename_create, link_document_type_filename_delete,
     link_document_type_filename_edit, link_document_type_filename_list,
     link_document_type_list, link_document_type_policies,
-    link_document_type_setup, link_document_update_page_count,
+    link_document_type_setup, link_document_pages_reset,
     link_document_version_download, link_document_version_list,
+    link_document_version_multiple_page_count_update,
+    link_document_version_page_count_update,
     link_document_version_return_document, link_document_version_return_list,
     link_document_version_revert, link_document_version_view,
     link_duplicated_document_list, link_duplicated_document_scan,
@@ -87,10 +89,10 @@ from .permissions import (
     permission_document_download, permission_document_edit,
     permission_document_new_version, permission_document_print,
     permission_document_properties_edit, permission_document_restore,
-    permission_document_trash, permission_document_type_delete,
+    permission_document_tools, permission_document_trash,
-    permission_document_type_edit, permission_document_type_view,
+    permission_document_type_delete, permission_document_type_edit,
-    permission_document_version_revert, permission_document_version_view,
+    permission_document_type_view, permission_document_version_revert,
-    permission_document_view
+    permission_document_version_view, permission_document_view,
 )
 # Just import to initialize the search models
 from .search import document_search, document_page_search  # NOQA
@@ -121,10 +123,11 @@ class DocumentsApp(MayanAppConfig):
         DeletedDocument = self.get_model(model_name='DeletedDocument')
         Document = self.get_model(model_name='Document')
         DocumentPage = self.get_model(model_name='DocumentPage')
-        DocumentPageResult = self.get_model(model_name='DocumentPageResult')
+        DocumentPageResult = self.get_model(model_name='DocumentVersionPageResult')
         DocumentType = self.get_model(model_name='DocumentType')
         DocumentTypeFilename = self.get_model(model_name='DocumentTypeFilename')
         DocumentVersion = self.get_model(model_name='DocumentVersion')
+        DocumentVersionPage = self.get_model(model_name='DocumentVersionPage')
         DuplicatedDocument = self.get_model(model_name='DuplicatedDocument')
 
         DynamicSerializerField.add_serializer(
@@ -190,13 +193,15 @@ class DocumentsApp(MayanAppConfig):
                 permission_acl_edit, permission_acl_view,
                 permission_document_delete, permission_document_download,
                 permission_document_edit, permission_document_new_version,
-                permission_document_print, permission_document_properties_edit,
+                permission_document_print,
-                permission_document_restore, permission_document_trash,
+                permission_document_properties_edit,
-                permission_document_version_revert,
+                permission_document_restore, permission_document_tools,
+                permission_document_trash, permission_document_version_revert,
                 permission_document_version_view, permission_document_view,
                 permission_events_view, permission_transformation_create,
                 permission_transformation_delete,
-                permission_transformation_edit, permission_transformation_view,
+                permission_transformation_edit,
+                permission_transformation_view,
             )
         )
 
@@ -224,13 +229,13 @@ class DocumentsApp(MayanAppConfig):
             model=Document, manager_name='passthrough'
         )
         ModelPermission.register_inheritance(
-            model=DocumentPage, related='document_version__document',
+            model=DocumentPage, related='document',
         )
         ModelPermission.register_manager(
             model=DocumentPage, manager_name='passthrough'
         )
         ModelPermission.register_inheritance(
-            model=DocumentPageResult, related='document_version__document',
+            model=DocumentPageResult, related='document',
         )
         ModelPermission.register_manager(
             model=DocumentPageResult, manager_name='passthrough'
@@ -241,6 +246,9 @@ class DocumentsApp(MayanAppConfig):
         ModelPermission.register_inheritance(
             model=DocumentVersion, related='document',
         )
+        ModelPermission.register_inheritance(
+            model=DocumentVersionPage, related='document_version',
+        )
 
         # Document and document page thumbnail widget
         document_page_thumbnail_widget = DocumentPageThumbnailWidget()
@@ -454,7 +462,7 @@ class DocumentsApp(MayanAppConfig):
                 link_document_quick_download, link_document_download,
                 link_document_clear_transformations,
                 link_document_clone_transformations,
-                link_document_update_page_count,
+                link_document_pages_reset,
             ), sources=(Document,)
         )
         menu_object.bind_links(
@@ -495,7 +503,7 @@ class DocumentsApp(MayanAppConfig):
                 link_document_multiple_favorites_remove,
                 link_document_multiple_clear_transformations,
                 link_document_multiple_trash, link_document_multiple_download,
-                link_document_multiple_update_page_count,
+                link_document_multiple_pages_reset,
                 link_document_multiple_document_type_edit,
             ), sources=(Document,)
         )
@@ -547,6 +555,17 @@ class DocumentsApp(MayanAppConfig):
                 link_document_version_return_list
             ), sources=(DocumentVersion,)
         )
+        menu_multi_item.bind_links(
+            links=(
+                link_document_version_multiple_page_count_update,
+            ), sources=(DocumentVersion,)
+        )
+        menu_object.bind_links(
+            links=(
+                link_document_version_page_count_update,
+            ), sources=(DocumentVersion,)
+        )
+
         menu_list_facet.bind_links(
             links=(link_document_version_view,), sources=(DocumentVersion,)
         )

mayan/apps/documents/dashboard_widgets.py

@@ -32,12 +32,12 @@ class DashboardWidgetDocumentPagesTotal(DashboardWidgetNumeric):
         AccessControlList = apps.get_model(
             app_label='acls', model_name='AccessControlList'
         )
-        DocumentPage = apps.get_model(
+        DocumentVersionPage = apps.get_model(
-            app_label='documents', model_name='DocumentPage'
+            app_label='documents', model_name='DocumentVersionPage'
         )
         self.count = AccessControlList.objects.restrict_queryset(
             permission=permission_document_view, user=request.user,
-            queryset=DocumentPage.objects.all()
+            queryset=DocumentVersionPage.objects.all()
         ).count()
         return super(DashboardWidgetDocumentPagesTotal, self).render(request)
 

mayan/apps/documents/icons.py

@@ -36,7 +36,10 @@ icon_document_edit = Icon(
 )
 icon_document = Icon(driver_name='fontawesome', symbol='book')
 icon_document_list = icon_document
-icon_document_page_count_update = Icon(
+icon_document_pages_reset = Icon(
+    driver_name='fontawesome', symbol='copy'
+)
+icon_document_version_page_count_update = Icon(
     driver_name='fontawesome', symbol='copy'
 )
 icon_document_preview = Icon(driver_name='fontawesome', symbol='eye')

mayan/apps/documents/links.py

@@ -168,12 +168,12 @@ link_document_quick_download = Link(
     permissions=(permission_document_download,), text=_('Quick download'),
     view='documents:document_download',
 )
-link_document_update_page_count = Link(
+link_document_pages_reset = Link(
     args='resolved_object.pk',
-    icon_class_path='mayan.apps.documents.icons.icon_document_page_count_update',
+    icon_class_path='mayan.apps.documents.icons.icon_document_pages_reset',
     permissions=(permission_document_tools,),
-    text=_('Recalculate page count'),
+    text=_('Reset pages'),
-    view='documents:document_update_page_count'
+    view='documents:document_pages_reset'
 )
 link_document_restore = Link(
     permissions=(permission_document_restore,),
@@ -217,10 +217,10 @@ link_document_multiple_download = Link(
     text=_('Advanced download'),
     view='documents:document_multiple_download_form'
 )
-link_document_multiple_update_page_count = Link(
+link_document_multiple_pages_reset = Link(
-    icon_class_path='mayan.apps.documents.icons.icon_document_page_count_update',
+    icon_class_path='mayan.apps.documents.icons.icon_document_pages_reset',
-    text=_('Recalculate page count'),
+    text=_('Reset pages'),
-    view='documents:document_multiple_update_page_count'
+    view='documents:document_multiple_pages_reset'
 )
 link_document_multiple_restore = Link(
     icon_class_path='mayan.apps.documents.icons.icon_trashed_document_restore',
@@ -246,6 +246,18 @@ link_document_version_return_list = Link(
     permissions=(permission_document_version_view,), text=_('Versions'),
     view='documents:document_version_list',
 )
+link_document_version_page_count_update = Link(
+    args='resolved_object.pk',
+    icon_class_path='mayan.apps.documents.icons.icon_document_version_page_count_update',
+    permissions=(permission_document_tools,),
+    text=_('Update page count'),
+    view='documents:document_version_page_count_update'
+)
+link_document_version_multiple_page_count_update = Link(
+    icon_class_path='mayan.apps.documents.icons.icon_document_version_page_count_update',
+    text=_('Update page count'),
+    view='documents:document_version_multiple_page_count_update'
+)
 link_document_version_view = Link(
     args='resolved_object.pk',
     icon_class_path='mayan.apps.documents.icons.icon_document_version_view',

mayan/apps/documents/literals.py

@@ -43,3 +43,5 @@ PAGE_RANGE_RANGE = 'range'
 PAGE_RANGE_CHOICES = (
     (PAGE_RANGE_ALL, _('All pages')), (PAGE_RANGE_RANGE, _('Page range'))
 )
+
+RETRY_DELAY_DOCUMENT_RESET_PAGES = 30

mayan/apps/documents/managers.py

@@ -28,15 +28,15 @@ class DocumentManager(models.Manager):
 
 class DocumentPageManager(models.Manager):
     def get_by_natural_key(self, page_number, document_version_natural_key):
-        DocumentVersion = apps.get_model(
+        Document = apps.get_model(
-            app_label='documents', model_name='DocumentVersion'
+            app_label='documents', model_name='Document'
         )
         try:
-            document_version = DocumentVersion.objects.get_by_natural_key(*document_version_natural_key)
+            document = Document.objects.get_by_natural_key(*document_version_natural_key)
-        except DocumentVersion.DoesNotExist:
+        except Document.DoesNotExist:
             raise self.model.DoesNotExist
 
-        return self.get(document_version__pk=document_version.pk, page_number=page_number)
+        return self.get(document__pk=document.pk, page_number=page_number)
 
     def get_queryset(self):
         return models.QuerySet(
@@ -124,6 +124,19 @@ class DocumentVersionManager(models.Manager):
         return self.get(document__pk=document.pk, checksum=checksum)
 
 
+class DocumentVersionPageManager(models.Manager):
+    def get_by_natural_key(self, page_number, document_version_natural_key):
+        DocumentVersion = apps.get_model(
+            app_label='documents', model_name='DocumentVersion'
+        )
+        try:
+            document_version = DocumentVersion.objects.get_by_natural_key(*document_version_natural_key)
+        except DocumentVersion.DoesNotExist:
+            raise self.model.DoesNotExist
+
+        return self.get(document_version__pk=document_version.pk, page_number=page_number)
+
+
 class DuplicatedDocumentManager(models.Manager):
     def clean_empty_duplicate_lists(self):
         self.filter(documents=None).delete()

mayan/apps/documents/migrations/0052_rename_document_page.py

@@ -0,0 +1,38 @@
+from __future__ import unicode_literals
+
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ('documents', '0051_documentpage_enabled'),
+    ]
+
+    operations = [
+        migrations.DeleteModel(
+            name='DocumentPageResult',
+        ),
+        migrations.RenameModel('DocumentPage', 'DocumentVersionPage'),
+        migrations.AlterField(
+            model_name='documentversionpage',
+            name='document_version',
+            field=models.ForeignKey(
+                on_delete=django.db.models.deletion.CASCADE,
+                related_name='pages', to='documents.DocumentVersion',
+                verbose_name='Document version'
+            ),
+        ),
+        migrations.RemoveField(
+            model_name='documentversionpage',
+            name='enabled',
+        ),
+        migrations.AlterModelOptions(
+            name='documentversionpage',
+            options={
+                'ordering': ('page_number',),
+                'verbose_name': 'Document version page',
+                'verbose_name_plural': 'Document version pages'
+            },
+        ),
+    ]

mayan/apps/documents/migrations/0053_create_document_page_and_result_models.py

@@ -0,0 +1,57 @@
+from __future__ import unicode_literals
+
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ('contenttypes', '0002_remove_content_type_name'),
+        ('documents', '0052_rename_document_page'),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name='DocumentPage',
+            fields=[
+                ('id', models.AutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('enabled', models.BooleanField(default=True, verbose_name='Enabled')),
+                ('page_number', models.PositiveIntegerField(blank=True, db_index=True, null=True, verbose_name='Page number')),
+                ('object_id', models.PositiveIntegerField()),
+                ('content_type', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, to='contenttypes.ContentType')),
+                ('document', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, related_name='pages', to='documents.Document', verbose_name='Document')),
+            ],
+            options={
+                'unique_together': set([('document', 'page_number')]),
+                'verbose_name': 'Document page',
+                'verbose_name_plural': 'Document pages',
+                'ordering': ('page_number',),
+            },
+        ),
+        migrations.CreateModel(
+            name='DocumentPageResult',
+            fields=[
+            ],
+            options={
+                'verbose_name': 'Document page result',
+                'verbose_name_plural': 'Document pages result',
+                'ordering': ('document', 'page_number'),
+                'proxy': True,
+                'indexes': [],
+            },
+            bases=('documents.documentpage',),
+        ),
+        migrations.CreateModel(
+            name='DocumentVersionPageResult',
+            fields=[
+            ],
+            options={
+                'verbose_name': 'Document version page',
+                'verbose_name_plural': 'Document version pages',
+                'ordering': ('document_version__document', 'page_number'),
+                'proxy': True,
+                'indexes': [],
+            },
+            bases=('documents.documentversionpage',),
+        ),
+    ]

mayan/apps/documents/migrations/0054_reset_document_pages.py

@@ -0,0 +1,62 @@
+from __future__ import unicode_literals
+
+from django.db import migrations
+
+
+def get_latest_version(document):
+    return document.versions.order_by('timestamp').last()
+
+
+def operation_reset_document_pages(apps, schema_editor):
+    Document = apps.get_model(app_label='documents', model_name='Document')
+    DocumentPage = apps.get_model(
+        app_label='documents', model_name='DocumentPage'
+    )
+
+    # Define inside the function to use the migration's apps instance
+    def pages_reset(document):
+        ContentType = apps.get_model('contenttypes', 'ContentType')
+        DocumentVersionPage = apps.get_model(
+            app_label='documents', model_name='DocumentVersionPage'
+        )
+
+        content_type = ContentType.objects.get_for_model(
+            model=DocumentVersionPage
+        )
+
+        for document_page in document.pages.all():
+            document_page.delete()
+
+        for version_page in get_latest_version(document=document).pages.all():
+            document_page = document.pages.create(
+                content_type=content_type,
+                page_number=version_page.page_number,
+                object_id=version_page.pk,
+            )
+
+    for document in Document.objects.using(schema_editor.connection.alias).all():
+        pages_reset(document=document)
+
+
+def operation_reset_document_pages_reverse(apps, schema_editor):
+    Document = apps.get_model(app_label='documents', model_name='Document')
+    DocumentPage = apps.get_model(
+        app_label='documents', model_name='DocumentPage'
+    )
+
+    for document in Document.objects.using(schema_editor.connection.alias).all():
+        for document_page in document.pages.all():
+            document_page.delete()
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ('documents', '0053_create_document_page_and_result_models'),
+    ]
+
+    operations = [
+        migrations.RunPython(
+            code=operation_reset_document_pages,
+            reverse_code=operation_reset_document_pages_reverse
+        ),
+    ]

mayan/apps/documents/models/__init__.py

@@ -2,4 +2,5 @@ from .document_models import *  # NOQA
 from .document_page_models import *  # NOQA
 from .document_type_models import *  # NOQA
 from .document_version_models import *  # NOQA
+from .document_version_page_models import *  # NOQA
 from .misc_models import *  # NOQA

mayan/apps/documents/models/document_models.py

@@ -5,9 +5,10 @@ import uuid
 
 from django.apps import apps
 from django.core.files import File
-from django.db import models
+from django.db import models, transaction
 from django.urls import reverse
 from django.utils.encoding import python_2_unicode_compatible
+from django.utils.functional import cached_property
 from django.utils.timezone import now
 from django.utils.translation import ugettext, ugettext_lazy as _
 
@@ -15,6 +16,7 @@ from ..events import (
     event_document_create, event_document_properties_edit,
     event_document_type_change,
 )
+from ..literals import DOCUMENT_IMAGES_CACHE_NAME
 from ..managers import DocumentManager, PassthroughManager, TrashCanManager
 from ..settings import setting_language
 from ..signals import post_document_type_change
@@ -102,6 +104,26 @@ class Document(models.Model):
         )
         return RecentDocument.objects.add_document_for_user(user, self)
 
+    @cached_property
+    def cache(self):
+        Cache = apps.get_model(app_label='file_caching', model_name='Cache')
+        return Cache.objects.get(name=DOCUMENT_IMAGES_CACHE_NAME)
+
+    @cached_property
+    def cache_partition(self):
+        partition, created = self.cache.partitions.get_or_create(
+            name='document-{}'.format(self.uuid)
+        )
+        return partition
+
+    @property
+    def checksum(self):
+        return self.latest_version.checksum
+
+    @property
+    def date_updated(self):
+        return self.latest_version.timestamp
+
     def delete(self, *args, **kwargs):
         to_trash = kwargs.pop('to_trash', True)
 
@@ -126,25 +148,37 @@ class Document(models.Model):
         else:
             return False
 
+    @property
+    def file_mime_encoding(self):
+        return self.latest_version.encoding
+
+    @property
+    def file_mimetype(self):
+        return self.latest_version.mimetype
+
     def get_absolute_url(self):
         return reverse(
             viewname='documents:document_preview', kwargs={'pk': self.pk}
         )
 
     def get_api_image_url(self, *args, **kwargs):
-        latest_version = self.latest_version
+        first_page = self.pages.first()
-        if latest_version:
+        if first_page:
-            return latest_version.get_api_image_url(*args, **kwargs)
+            return first_page.get_api_image_url(*args, **kwargs)
 
     @property
     def is_in_trash(self):
         return self.in_trash
 
+    @property
+    def latest_version(self):
+        return self.versions.order_by('timestamp').last()
+
     def natural_key(self):
         return (self.uuid,)
     natural_key.dependencies = ['documents.DocumentType']
 
-    def new_version(self, file_object, comment=None, _user=None):
+    def new_version(self, file_object, append_pages=False, comment=None, _user=None):
         logger.info('Creating new document version for document: %s', self)
         DocumentVersion = apps.get_model(
             app_label='documents', model_name='DocumentVersion'
@@ -153,9 +187,10 @@ class Document(models.Model):
         document_version = DocumentVersion(
             document=self, comment=comment or '', file=File(file_object)
         )
-        document_version.save(_user=_user)
+        document_version.save(append_pages=append_pages, _user=_user)
 
         logger.info('New document version queued for document: %s', self)
+
         return document_version
 
     def open(self, *args, **kwargs):
@@ -165,6 +200,34 @@ class Document(models.Model):
         """
         return self.latest_version.open(*args, **kwargs)
 
+    @property
+    def page_count(self):
+        return self.pages.count()
+
+    @property
+    def pages(self):
+        return self.pages.all()
+
+    @property
+    def pages_all(self):
+        DocumentPage = apps.get_model(
+            app_label='documents', model_name='DocumentPage'
+        )
+        return DocumentPage.passthrough.filter(document=self)
+
+    def pages_reset(self, update_page_count=True):
+        with transaction.atomic():
+            for page in self.pages.all():
+                page.delete()
+
+            if update_page_count:
+                self.latest_version.update_page_count()
+
+            for version_page in self.latest_version.pages.all():
+                document_page = self.pages.create(
+                    content_object = version_page
+                )
+
     def restore(self):
         self.in_trash = False
         self.save()
@@ -209,53 +272,3 @@ class Document(models.Model):
     @property
     def size(self):
         return self.latest_version.size
-
-    # Compatibility methods
-
-    @property
-    def checksum(self):
-        return self.latest_version.checksum
-
-    @property
-    def date_updated(self):
-        return self.latest_version.timestamp
-
-    @property
-    def file_mime_encoding(self):
-        return self.latest_version.encoding
-
-    @property
-    def file_mimetype(self):
-        return self.latest_version.mimetype
-
-    @property
-    def latest_version(self):
-        return self.versions.order_by('timestamp').last()
-
-    @property
-    def page_count(self):
-        return self.latest_version.page_count
-
-    @property
-    def pages_all(self):
-        try:
-            return self.latest_version.pages_all
-        except AttributeError:
-            # Document has no version yet
-            DocumentPage = apps.get_model(
-                app_label='documents', model_name='DocumentPage'
-            )
-
-            return DocumentPage.objects.none()
-
-    @property
-    def pages(self):
-        try:
-            return self.latest_version.pages
-        except AttributeError:
-            # Document has no version yet
-            DocumentPage = apps.get_model(
-                app_label='documents', model_name='DocumentPage'
-            )
-
-            return DocumentPage.objects.none()

mayan/apps/documents/models/document_page_models.py

@@ -4,14 +4,16 @@ import logging
 
 from furl import furl
 
+from django.contrib.contenttypes.fields import GenericForeignKey
+from django.contrib.contenttypes.models import ContentType
 from django.db import models
+from django.db.models import Max
 from django.urls import reverse
 from django.utils.encoding import force_text, python_2_unicode_compatible
 from django.utils.functional import cached_property
 from django.utils.translation import ugettext_lazy as _
 
 from mayan.apps.converter.literals import DEFAULT_ZOOM_LEVEL, DEFAULT_ROTATION
-
 from mayan.apps.converter.models import LayerTransformation
 from mayan.apps.converter.transformations import (
     BaseTransformation, TransformationResize, TransformationRotate,
@@ -26,7 +28,7 @@ from ..settings import (
     setting_zoom_min_level
 )
 
-from .document_version_models import DocumentVersion
+from .document_models import Document
 
 __all__ = ('DocumentPage', 'DocumentPageResult')
 logger = logging.getLogger(__name__)
@@ -35,16 +37,22 @@ logger = logging.getLogger(__name__)
 @python_2_unicode_compatible
 class DocumentPage(models.Model):
     """
-    Model that describes a document version page
+    Model that describes a document page
     """
-    document_version = models.ForeignKey(
+    document = models.ForeignKey(
-        on_delete=models.CASCADE, related_name='version_pages', to=DocumentVersion,
+        on_delete=models.CASCADE, related_name='pages', to=Document,
-        verbose_name=_('Document version')
+        verbose_name=_('Document')
     )
     enabled = models.BooleanField(default=True, verbose_name=_('Enabled'))
     page_number = models.PositiveIntegerField(
-        db_index=True, default=1, editable=False,
+        db_index=True, blank=True, null=True, verbose_name=_('Page number')
-        verbose_name=_('Page number')
+    )
+    content_type = models.ForeignKey(
+        on_delete=models.CASCADE, to=ContentType
+    )
+    object_id = models.PositiveIntegerField()
+    content_object = GenericForeignKey(
+        ct_field='content_type', fk_field='object_id'
     )
 
     objects = DocumentPageManager()
@@ -52,6 +60,7 @@ class DocumentPage(models.Model):
 
     class Meta:
         ordering = ('page_number',)
+        unique_together = ('document', 'page_number')
         verbose_name = _('Document page')
         verbose_name_plural = _('Document pages')
 
@@ -60,7 +69,7 @@ class DocumentPage(models.Model):
 
     @cached_property
     def cache_partition(self):
-        partition, created = self.document_version.cache.partitions.get_or_create(
+        partition, created = self.document.cache.partitions.get_or_create(
             name=self.uuid
         )
         return partition
@@ -69,19 +78,15 @@ class DocumentPage(models.Model):
         self.cache_partition.delete()
         super(DocumentPage, self).delete(*args, **kwargs)
 
-    def detect_orientation(self):
+    #def detect_orientation(self):
-        with self.document_version.open() as file_object:
+    #    with self.document_version.open() as file_object:
-            converter = get_converter_class()(
+    #        converter = get_converter_class()(
-                file_object=file_object,
+    #            file_object=file_object,
-                mime_type=self.document_version.mimetype
+    #            mime_type=self.document_version.mimetype
-            )
+    #        )
-            return converter.detect_orientation(
+    #        return converter.detect_orientation(
-                page_number=self.page_number
+    #            page_number=self.page_number
-            )
+    #        )
-
-    @property
-    def document(self):
-        return self.document_version.document
 
     def generate_image(self, user=None, **kwargs):
         transformation_list = self.get_combined_transformation_list(user=user, **kwargs)
@@ -90,7 +95,7 @@ class DocumentPage(models.Model):
         # Check is transformed image is available
         logger.debug('transformations cache filename: %s', combined_cache_filename)
 
-        if not setting_disable_transformed_image_cache.value and self.cache_partition.get_file(filename=combined_cache_filename):
+        if self.cache_partition.get_file(filename=combined_cache_filename):
             logger.debug(
                 'transformations cache file "%s" found', combined_cache_filename
             )
@@ -128,8 +133,7 @@ class DocumentPage(models.Model):
         final_url.args = kwargs
         final_url.path = reverse(
             viewname='rest_api:documentpage-image', kwargs={
-                'pk': self.document.pk, 'version_pk': self.document_version.pk,
+                'pk': self.document.pk, 'page_pk': self.pk
-                'page_pk': self.pk
             }
         )
         final_url.args['_hash'] = transformations_hash
@@ -190,12 +194,12 @@ class DocumentPage(models.Model):
         return transformation_list
 
     def get_image(self, transformations=None):
-        cache_filename = 'base_image'
+        cache_filename = 'document_page'
         logger.debug('Page cache filename: %s', cache_filename)
 
         cache_file = self.cache_partition.get_file(filename=cache_filename)
 
-        if not setting_disable_base_image_cache.value and cache_file:
+        if cache_file:
             logger.debug('Page cache file "%s" found', cache_filename)
 
             with cache_file.open() as file_object:
@@ -216,14 +220,25 @@ class DocumentPage(models.Model):
             logger.debug('Page cache file "%s" not found', cache_filename)
 
             try:
-                with self.document_version.get_intermediate_file() as file_object:
+                #with self.document_version.get_intermediate_file() as file_object:
+                #Render or get cached document version page
+
+                #self.content_object.generate_image()
+                self.content_object.get_image()
+                cache_filename = 'base_image'
+                cache_file = self.content_object.cache_partition.get_file(filename=cache_filename)
+
+                with cache_file.open() as file_object:
                     converter = get_converter_class()(
                         file_object=file_object
                     )
-                    converter.seek_page(page_number=self.page_number - 1)
+                    converter.seek_page(page_number=0)
+                    #self.page_number - 1)
 
                     page_image = converter.get_page()
 
+                    cache_filename = 'document_page'
+
                     # Since open "wb+" doesn't create files, create it explicitly
                     with self.cache_partition.create_file(filename=cache_filename) as file_object:
                         file_object.write(page_image.getvalue())
@@ -241,28 +256,39 @@ class DocumentPage(models.Model):
                 )
                 raise
 
+    def get_label(self):
+        return _(
+            'Page %(page_number)d out of %(total_pages)d of %(document)s'
+        ) % {
+            'document': force_text(self.document),
+            'page_number': self.page_number,
+            'total_pages': self.document.pages_all.count()
+        }
+    get_label.short_description = _('Label')
+
     @property
     def is_in_trash(self):
         return self.document.is_in_trash
 
-    def get_label(self):
-        return _(
-            'Page %(page_num)d out of %(total_pages)d of %(document)s'
-        ) % {
-            'document': force_text(self.document),
-            'page_num': self.page_number,
-            'total_pages': self.document_version.pages_all.count()
-        }
-    get_label.short_description = _('Label')
-
     def natural_key(self):
-        return (self.page_number, self.document_version.natural_key())
+        return (self.page_number, self.document.natural_key())
-    natural_key.dependencies = ['documents.DocumentVersion']
+    natural_key.dependencies = ['documents.Document']
+
+    def save(self, *args, **kwargs):
+        if not self.page_number:
+            last_page_number = DocumentPage.objects.filter(
+                document=self.document
+            ).aggregate(Max('page_number'))['page_number__max']
+            if last_page_number is not None:
+                self.page_number = last_page_number + 1
+            else:
+                self.page_number = 1
+        super(DocumentPage, self).save(*args, **kwargs)
 
     @property
     def siblings(self):
         return DocumentPage.objects.filter(
-            document_version=self.document_version
+            document=self.document
         )
 
     @property
@@ -271,12 +297,12 @@ class DocumentPage(models.Model):
         Make cache UUID a mix of version ID and page ID to avoid using stale
         images
         """
-        return '{}-{}'.format(self.document_version.uuid, self.pk)
+        return '{}-{}'.format(self.document.uuid, self.pk)
 
 
 class DocumentPageResult(DocumentPage):
     class Meta:
-        ordering = ('document_version__document', 'page_number')
+        ordering = ('document', 'page_number')
         proxy = True
-        verbose_name = _('Document page')
+        verbose_name = _('Document page result')
-        verbose_name_plural = _('Document pages')
+        verbose_name_plural = _('Document pages result')

mayan/apps/documents/models/document_version_models.py

@@ -246,23 +246,12 @@ class DocumentVersion(models.Model):
 
             return result
 
-    @property
+    #@property
-    def pages_all(self):
+    #def page_count(self):
-        DocumentPage = apps.get_model(
+    #    """
-            app_label='documents', model_name='DocumentPage'
+    #    The number of pages that the document posses.
-        )
+    #    """
-        return DocumentPage.passthrough.filter(document_version=self)
+    #    return self.pages.count()
-
-    @property
-    def pages(self):
-        return self.version_pages.all()
-
-    @property
-    def page_count(self):
-        """
-        The number of pages that the document posses.
-        """
-        return self.pages.count()
 
     def revert(self, _user=None):
         """
@@ -285,6 +274,7 @@ class DocumentVersion(models.Model):
         Overloaded save method that updates the document version's checksum,
         mimetype, and page count when created
         """
+        append_pages = kwargs.pop('append_pages', False)
         user = kwargs.pop('_user', None)
         new_document_version = not self.pk
 
@@ -304,7 +294,7 @@ class DocumentVersion(models.Model):
                     # Only do this for new documents
                     self.update_checksum(save=False)
                     self.update_mimetype(save=False)
-                    self.save()
+                    self.save(append_pages=append_pages, _user=user)
                     self.update_page_count(save=False)
                     if setting_fix_orientation.value:
                         self.fix_orientation()
@@ -337,6 +327,14 @@ class DocumentVersion(models.Model):
                         sender=Document, instance=self.document
                     )
 
+                if append_pages:
+                    for version_page in self.pages.all():
+                        self.document.pages.create(
+                            content_object = version_page
+                        )
+                else:
+                    self.document.pages_reset(update_page_count=False)
+
     def save_to_file(self, file_object):
         """
         Save a copy of the document from the document storage backend
@@ -410,7 +408,7 @@ class DocumentVersion(models.Model):
             pass
         else:
             DocumentPage = apps.get_model(
-                app_label='documents', model_name='DocumentPage'
+                app_label='documents', model_name='DocumentVersionPage'
             )
 
             with transaction.atomic():

mayan/apps/documents/models/document_version_page_models.py

@@ -0,0 +1,282 @@
+from __future__ import absolute_import, unicode_literals
+
+import logging
+
+from furl import furl
+
+from django.db import models
+from django.urls import reverse
+from django.utils.encoding import force_text, python_2_unicode_compatible
+from django.utils.functional import cached_property
+from django.utils.translation import ugettext_lazy as _
+
+from mayan.apps.converter.literals import DEFAULT_ZOOM_LEVEL, DEFAULT_ROTATION
+
+from mayan.apps.converter.models import LayerTransformation
+from mayan.apps.converter.transformations import (
+    BaseTransformation, TransformationResize, TransformationRotate,
+    TransformationZoom
+)
+from mayan.apps.converter.utils import get_converter_class
+
+from ..managers import DocumentVersionPageManager
+from ..settings import (
+    setting_disable_base_image_cache, setting_disable_transformed_image_cache,
+    setting_display_width, setting_display_height, setting_zoom_max_level,
+    setting_zoom_min_level
+)
+
+from .document_version_models import DocumentVersion
+
+__all__ = ('DocumentVersionPage', 'DocumentVersionPageResult')
+logger = logging.getLogger(__name__)
+
+
+@python_2_unicode_compatible
+class DocumentVersionPage(models.Model):
+    """
+    Model that describes a document version page
+    """
+    document_version = models.ForeignKey(
+        on_delete=models.CASCADE, related_name='pages', to=DocumentVersion,
+        verbose_name=_('Document version')
+    )
+    page_number = models.PositiveIntegerField(
+        db_index=True, default=1, editable=False,
+        verbose_name=_('Page number')
+    )
+
+    objects = DocumentVersionPageManager()
+
+    class Meta:
+        ordering = ('page_number',)
+        verbose_name = _('Document version page')
+        verbose_name_plural = _('Document version pages')
+
+    def __str__(self):
+        return self.get_label()
+
+    @cached_property
+    def cache_partition(self):
+        partition, created = self.document_version.cache.partitions.get_or_create(
+            name=self.uuid
+        )
+        return partition
+
+    def delete(self, *args, **kwargs):
+        self.cache_partition.delete()
+        super(DocumentVersionPage, self).delete(*args, **kwargs)
+
+    #def detect_orientation(self):
+    #    with self.document_version.open() as file_object:
+    #        converter = get_converter_class()(
+    #            file_object=file_object,
+    #            mime_type=self.document_version.mimetype
+    #        )
+    #        return converter.detect_orientation(
+    #            page_number=self.page_number
+    #        )
+
+    @property
+    def document(self):
+        return self.document_version.document
+
+    def generate_image(self, user=None, **kwargs):
+        transformation_list = self.get_combined_transformation_list(user=user, **kwargs)
+        combined_cache_filename = BaseTransformation.combine(transformation_list)
+
+        # Check is transformed image is available
+        logger.debug('transformations cache filename: %s', combined_cache_filename)
+
+        if self.cache_partition.get_file(filename=combined_cache_filename):
+            logger.debug(
+                'transformations cache file "%s" found', combined_cache_filename
+            )
+        else:
+            logger.debug(
+                'transformations cache file "%s" not found', combined_cache_filename
+            )
+            image = self.get_image(transformations=transformation_list)
+            with self.cache_partition.create_file(filename=combined_cache_filename) as file_object:
+                file_object.write(image.getvalue())
+
+        return combined_cache_filename
+
+    #def get_absolute_url(self):
+    #    return reverse(
+    #        viewname='documents:document_version_page_view', kwargs={
+    #            'pk': self.pk
+    #        }
+    #    )
+
+    def get_api_image_url(self, *args, **kwargs):
+        """
+        Create an unique URL combining:
+        - the page's image URL
+        - the interactive argument
+        - a hash from the server side and interactive transformations
+        The purpose of this unique URL is to allow client side caching
+        if document page images.
+        """
+        transformations_hash = BaseTransformation.combine(
+            self.get_combined_transformation_list(*args, **kwargs)
+        )
+
+        kwargs.pop('transformations', None)
+
+        final_url = furl()
+        final_url.args = kwargs
+        final_url.path = reverse(
+            viewname='rest_api:documentversionpage-image', kwargs={
+                'pk': self.document.pk, 'version_pk': self.document_version.pk,
+                'page_pk': self.pk
+            }
+        )
+        final_url.args['_hash'] = transformations_hash
+
+        return final_url.tostr()
+
+    def get_combined_transformation_list(self, user=None, *args, **kwargs):
+        """
+        Return a list of transformation containing the server side
+        document page transformation as well as tranformations created
+        from the arguments as transient interactive transformation.
+        """
+        # Convert arguments into transformations
+        transformations = kwargs.get('transformations', [])
+
+        # Set sensible defaults if the argument is not specified or if the
+        # argument is None
+        width = kwargs.get('width', setting_display_width.value) or setting_display_width.value
+        height = kwargs.get('height', setting_display_height.value) or setting_display_height.value
+        rotation = kwargs.get('rotation', DEFAULT_ROTATION) or DEFAULT_ROTATION
+        zoom_level = kwargs.get('zoom', DEFAULT_ZOOM_LEVEL) or DEFAULT_ZOOM_LEVEL
+
+        if zoom_level < setting_zoom_min_level.value:
+            zoom_level = setting_zoom_min_level.value
+
+        if zoom_level > setting_zoom_max_level.value:
+            zoom_level = setting_zoom_max_level.value
+
+        # Generate transformation hash
+        transformation_list = []
+
+        maximum_layer_order = kwargs.get('maximum_layer_order', None)
+
+        # Stored transformations first
+        for stored_transformation in LayerTransformation.objects.get_for_object(
+            self, maximum_layer_order=maximum_layer_order, as_classes=True,
+            user=user
+        ):
+            transformation_list.append(stored_transformation)
+
+        # Interactive transformations second
+        for transformation in transformations:
+            transformation_list.append(transformation)
+
+        if rotation:
+            transformation_list.append(
+                TransformationRotate(degrees=rotation)
+            )
+
+        if width:
+            transformation_list.append(
+                TransformationResize(width=width, height=height)
+            )
+
+        if zoom_level:
+            transformation_list.append(TransformationZoom(percent=zoom_level))
+
+        return transformation_list
+
+    def get_image(self, transformations=None):
+        cache_filename = 'base_image'
+        logger.debug('Page cache filename: %s', cache_filename)
+
+        cache_file = self.cache_partition.get_file(filename=cache_filename)
+
+        if cache_file:
+            logger.debug('Page cache file "%s" found', cache_filename)
+
+            with cache_file.open() as file_object:
+                converter = get_converter_class()(
+                    file_object=file_object
+                )
+
+                converter.seek_page(page_number=0)
+
+                # This code is also repeated below to allow using a context
+                # manager with cache_file.open and close it automatically.
+                # Apply runtime transformations
+                for transformation in transformations or []:
+                    converter.transform(transformation=transformation)
+
+                return converter.get_page()
+        else:
+            logger.debug('Page cache file "%s" not found', cache_filename)
+
+            try:
+                with self.document_version.get_intermediate_file() as file_object:
+                    converter = get_converter_class()(
+                        file_object=file_object
+                    )
+                    converter.seek_page(page_number=self.page_number - 1)
+
+                    page_image = converter.get_page()
+
+                    # Since open "wb+" doesn't create files, create it explicitly
+                    with self.cache_partition.create_file(filename=cache_filename) as file_object:
+                        file_object.write(page_image.getvalue())
+
+                    # Apply runtime transformations
+                    for transformation in transformations or []:
+                        converter.transform(transformation=transformation)
+
+                    return converter.get_page()
+            except Exception as exception:
+                # Cleanup in case of error
+                logger.error(
+                    'Error creating page cache file "%s"; %s',
+                    cache_filename, exception
+                )
+                raise
+
+    #@property
+    #def is_in_trash(self):
+    #    return self.document.is_in_trash
+
+    def get_label(self):
+        return _(
+            'Version page %(page_number)d out of %(total_pages)d of %(document)s'
+        ) % {
+            'document': force_text(self.document),
+            'page_number': self.page_number,
+            'total_pages': self.document_version.pages.count()
+        }
+    get_label.short_description = _('Label')
+
+    def natural_key(self):
+        return (self.page_number, self.document_version.natural_key())
+    natural_key.dependencies = ['documents.DocumentVersion']
+
+    @property
+    def siblings(self):
+        return DocumentVersionPage.objects.filter(
+            document_version=self.document_version
+        )
+
+    @property
+    def uuid(self):
+        """
+        Make cache UUID a mix of version ID and page ID to avoid using stale
+        images
+        """
+        return '{}-{}'.format(self.document_version.uuid, self.pk)
+
+
+class DocumentVersionPageResult(DocumentVersionPage):
+    class Meta:
+        ordering = ('document_version__document', 'page_number')
+        proxy = True
+        verbose_name = _('Document version page')
+        verbose_name_plural = _('Document version pages')

mayan/apps/documents/queues.py

@@ -30,6 +30,10 @@ queue_converter.add_task_type(
     dotted_path='mayan.apps.documents.tasks.task_generate_document_page_image',
     label=_('Generate document page image')
 )
+queue_converter.add_task_type(
+    dotted_path='mayan.apps.documents.tasks.task_generate_document_version_page_image',
+    label=_('Generate document version page image')
+)
 
 queue_documents.add_task_type(
     dotted_path='mayan.apps.documents.tasks.task_delete_document',
@@ -66,6 +70,10 @@ queue_tools.add_task_type(
     label=_('Duplicated document scan')
 )
 
+queue_uploads.add_task_type(
+    dotted_path='mayan.apps.documents.tasks.task_document_pages_reset',
+    label=_('Reset document pages')
+)
 queue_uploads.add_task_type(
     dotted_path='mayan.apps.documents.tasks.task_update_page_count',
     label=_('Update document page count')

mayan/apps/documents/search.py

@@ -17,12 +17,20 @@ def transformation_format_uuid(term_string):
         return term_string
 
 
-def get_queryset_page_search_queryset():
+def get_queryset_document_page_search_queryset():
     # Ignore documents in trash can
     DocumentPage = apps.get_model(
         app_label='documents', model_name='DocumentPage'
     )
-    return DocumentPage.objects.filter(document_version__document__in_trash=False)
+    return DocumentPage.objects.filter(document__in_trash=False)
+
+
+def get_queryset_document_version_page_search_queryset():
+    # Ignore documents in trash can
+    DocumentVersionPage = apps.get_model(
+        app_label='documents', model_name='DocumentVersionPage'
+    )
+    return DocumentVersionPage.objects.filter(document_version__document__in_trash=False)
 
 
 document_search = SearchModel(
@@ -30,7 +38,6 @@ document_search = SearchModel(
     model_name='Document', permission=permission_document_view,
     serializer_path='mayan.apps.documents.serializers.DocumentSerializer'
 )
-
 document_search.add_model_field(
     field='document_type__label', label=_('Document type')
 )
@@ -50,24 +57,49 @@ document_search.add_model_field(
 document_page_search = SearchModel(
     app_label='documents', list_mode=LIST_MODE_CHOICE_ITEM,
     model_name='DocumentPage', permission=permission_document_view,
-    queryset=get_queryset_page_search_queryset,
+    queryset=get_queryset_document_page_search_queryset,
     serializer_path='mayan.apps.documents.serializers.DocumentPageSerializer'
 )
 
+document_version_page_search = SearchModel(
+    app_label='documents', list_mode=LIST_MODE_CHOICE_ITEM,
+    model_name='DocumentVersionPage', permission=permission_document_view,
+    queryset=get_queryset_document_version_page_search_queryset,
+    serializer_path='mayan.apps.documents.serializers.DocumentVersionPageSerializer'
+)
+
 document_page_search.add_model_field(
-    field='document_version__document__document_type__label',
+    field='document__document_type__label',
     label=_('Document type')
 )
 document_page_search.add_model_field(
-    field='document_version__document__versions__mimetype',
+    field='document__versions__mimetype',
     label=_('MIME type')
 )
 document_page_search.add_model_field(
+    field='document__label', label=_('Label')
+)
+document_page_search.add_model_field(
+    field='document__description', label=_('Description')
+)
+document_page_search.add_model_field(
+    field='document__versions__checksum', label=_('Checksum')
+)
+
+document_version_page_search.add_model_field(
+    field='document_version__document__document_type__label',
+    label=_('Document type')
+)
+document_version_page_search.add_model_field(
+    field='document_version__document__versions__mimetype',
+    label=_('MIME type')
+)
+document_version_page_search.add_model_field(
     field='document_version__document__label', label=_('Label')
 )
-document_page_search.add_model_field(
+document_version_page_search.add_model_field(
     field='document_version__document__description', label=_('Description')
 )
-document_page_search.add_model_field(
+document_version_page_search.add_model_field(
     field='document_version__checksum', label=_('Checksum')
 )

mayan/apps/documents/serializers.py

@@ -8,42 +8,40 @@ from rest_framework.reverse import reverse
 from mayan.apps.common.models import SharedUploadedFile
 
 from .models import (
-    Document, DocumentVersion, DocumentPage, DocumentType,
+    Document, DocumentPage, DocumentType, DocumentTypeFilename,
-    DocumentTypeFilename, RecentDocument
+    DocumentVersion, DocumentVersionPage, RecentDocument
 )
 from .settings import setting_language
 from .tasks import task_upload_new_version
 
 
 class DocumentPageSerializer(serializers.HyperlinkedModelSerializer):
-    document_version_url = serializers.SerializerMethodField()
+    document_url = serializers.SerializerMethodField()
     image_url = serializers.SerializerMethodField()
     url = serializers.SerializerMethodField()
 
     class Meta:
-        fields = ('document_version_url', 'image_url', 'page_number', 'url')
+        fields = ('document_url', 'image_url', 'page_number', 'url')
         model = DocumentPage
 
-    def get_document_version_url(self, instance):
+    def get_document_url(self, instance):
         return reverse(
-            viewname='rest_api:documentversion-detail', args=(
+            viewname='rest_api:document-detail', args=(
-                instance.document.pk, instance.document_version.pk,
+                instance.document.pk,
             ), request=self.context['request'], format=self.context['format']
         )
 
     def get_image_url(self, instance):
         return reverse(
             viewname='rest_api:documentpage-image', args=(
-                instance.document.pk, instance.document_version.pk,
+                instance.document.pk, instance.pk,
-                instance.pk,
             ), request=self.context['request'], format=self.context['format']
         )
 
     def get_url(self, instance):
         return reverse(
             viewname='rest_api:documentpage-detail', args=(
-                instance.document.pk, instance.document_version.pk,
+                instance.document.pk, instance.pk,
-                instance.pk,
             ), request=self.context['request'], format=self.context['format']
         )
 
@@ -97,6 +95,39 @@ class WritableDocumentTypeSerializer(serializers.ModelSerializer):
         return obj.documents.count()
 
 
+class DocumentVersionPageSerializer(serializers.HyperlinkedModelSerializer):
+    document_version_url = serializers.SerializerMethodField()
+    image_url = serializers.SerializerMethodField()
+    url = serializers.SerializerMethodField()
+
+    class Meta:
+        fields = ('document_version_url', 'image_url', 'page_number', 'url')
+        model = DocumentVersionPage
+
+    def get_document_version_url(self, instance):
+        return reverse(
+            viewname='rest_api:documentversion-detail', args=(
+                instance.document.pk, instance.document_version.pk,
+            ), request=self.context['request'], format=self.context['format']
+        )
+
+    def get_image_url(self, instance):
+        return reverse(
+            viewname='rest_api:documentversionpage-image', args=(
+                instance.document.pk, instance.document_version.pk,
+                instance.pk,
+            ), request=self.context['request'], format=self.context['format']
+        )
+
+    def get_url(self, instance):
+        return reverse(
+            viewname='rest_api:documentversionpage-detail', args=(
+                instance.document.pk, instance.document_version.pk,
+                instance.pk,
+            ), request=self.context['request'], format=self.context['format']
+        )
+
+
 class DocumentVersionSerializer(serializers.HyperlinkedModelSerializer):
     document_url = serializers.SerializerMethodField()
     download_url = serializers.SerializerMethodField()

mayan/apps/documents/statistics.py

@@ -41,7 +41,7 @@ def new_documents_per_month():
 
 def new_document_pages_per_month():
     DocumentPage = apps.get_model(
-        app_label='documents', model_name='DocumentPage'
+        app_label='documents', model_name='DocumentVersionPage'
     )
 
     qss = qsstats.QuerySetStats(
@@ -106,7 +106,7 @@ def new_document_pages_this_month(user=None):
         app_label='acls', model_name='AccessControlList'
     )
     DocumentPage = apps.get_model(
-        app_label='documents', model_name='DocumentPage'
+        app_label='documents', model_name='DocumentVersionPage'
     )
 
     queryset = DocumentPage.objects.all()
@@ -195,7 +195,7 @@ def total_document_version_per_month():
 
 def total_document_page_per_month():
     DocumentPage = apps.get_model(
-        app_label='documents', model_name='DocumentPage'
+        app_label='documents', model_name='DocumentVersionPage'
     )
 
     qss = qsstats.QuerySetStats(

mayan/apps/documents/tasks.py

@@ -9,8 +9,8 @@ from django.db import OperationalError
 from mayan.celery import app
 
 from .literals import (
-    UPDATE_PAGE_COUNT_RETRY_DELAY, UPLOAD_NEW_DOCUMENT_RETRY_DELAY,
+    RETRY_DELAY_DOCUMENT_RESET_PAGES, UPDATE_PAGE_COUNT_RETRY_DELAY,
-    UPLOAD_NEW_VERSION_RETRY_DELAY
+    UPLOAD_NEW_DOCUMENT_RETRY_DELAY,
 )
 
 logger = logging.getLogger(__name__)
@@ -65,6 +65,25 @@ def task_delete_stubs():
     logger.info(msg='Finshed')
 
 
+@app.task(bind=True, default_retry_delay=RETRY_DELAY_DOCUMENT_RESET_PAGES, ignore_result=True)
+def task_document_pages_reset(self, document_id):
+    Document = apps.get_model(
+        app_label='documents', model_name='Document'
+    )
+
+    document = Document.objects.get(pk=document_id)
+
+    try:
+        document.pages_reset()
+    except OperationalError as exception:
+        logger.warning(
+            'Operational error during attempt to reset pages for '
+            'document: %s; %s. Retrying.', document,
+            exception
+        )
+        raise self.retry(exc=exception)
+
+
 @app.task()
 def task_generate_document_page_image(document_page_id, user_id=None, **kwargs):
     DocumentPage = apps.get_model(
@@ -81,6 +100,22 @@ def task_generate_document_page_image(document_page_id, user_id=None, **kwargs):
     return document_page.generate_image(user=user, **kwargs)
 
 
+@app.task()
+def task_generate_document_version_page_image(document_version_page_id, user_id=None, **kwargs):
+    DocumentVersionPage = apps.get_model(
+        app_label='documents', model_name='DocumentVersionPage'
+    )
+    User = get_user_model()
+
+    if user_id:
+        user = User.objects.get(pk=user_id)
+    else:
+        user = None
+
+    document_version_page = DocumentVersionPage.objects.get(pk=document_version_page_id)
+    return document_version_page.generate_image(user=user, **kwargs)
+
+
 @app.task(ignore_result=True)
 def task_scan_duplicates_all():
     DuplicatedDocument = apps.get_model(
@@ -177,7 +212,7 @@ def task_upload_new_document(self, document_type_id, shared_uploaded_file_id):
 
 
 @app.task(bind=True, default_retry_delay=UPLOAD_NEW_VERSION_RETRY_DELAY, ignore_result=True)
-def task_upload_new_version(self, document_id, shared_uploaded_file_id, user_id, comment=None):
+def task_upload_new_version(self, document_id, shared_uploaded_file_id, user_id, append_pages=False, comment=None):
     SharedUploadedFile = apps.get_model(
         app_label='common', model_name='SharedUploadedFile'
     )
@@ -212,7 +247,7 @@ def task_upload_new_version(self, document_id, shared_uploaded_file_id, user_id,
             document=document, comment=comment or '', file=file_object
         )
         try:
-            document_version.save(_user=user)
+            document_version.save(append_pages=append_pages, _user=user)
         except Warning as warning:
             # New document version are blocked
             logger.info(

mayan/apps/documents/tests/mixins.py

@@ -69,6 +69,8 @@ class DocumentTestMixin(object):
 
         self.test_document = document
         self.test_documents.append(document)
+        self.test_document_version = document.latest_version
+        self.test_document_page = document.pages_all.first()
 
 
 class DocumentTypeViewTestMixin(object):
@@ -148,6 +150,26 @@ class DocumentVersionTestMixin(object):
             )
 
 
+class DocumentVersionViewTestMixin(object):
+    def _request_document_version_list_view(self):
+        return self.get(
+            viewname='documents:document_version_list',
+            kwargs={'pk': self.test_document.pk}
+        )
+
+    def _request_document_version_revert_view(self, document_version):
+        return self.post(
+            viewname='documents:document_version_revert',
+            kwargs={'pk': document_version.pk}
+        )
+
+    def _request_test_document_version_page_count_update_view(self):
+        return self.post(
+            viewname='documents:document_version_page_count_update',
+            kwargs={'pk': self.test_document_version.pk}
+        )
+
+
 class DocumentViewTestMixin(object):
     def _request_document_properties_view(self):
         return self.get(
@@ -200,6 +222,12 @@ class DocumentViewTestMixin(object):
             data={'id_list': self.test_document.pk}
         )
 
+    def _request_document_pages_reset_view(self):
+        return self.post(
+            viewname='documents:document_pages_reset',
+            kwargs={'pk': self.test_document.pk}
+        )
+
     def _request_document_version_download(self, data=None):
         data = data or {}
         return self.get(
@@ -208,18 +236,6 @@ class DocumentViewTestMixin(object):
             }, data=data
         )
 
-    def _request_document_update_page_count_view(self):
-        return self.post(
-            viewname='documents:document_update_page_count',
-            kwargs={'pk': self.test_document.pk}
-        )
-
-    def _request_document_multiple_update_page_count_view(self):
-        return self.post(
-            viewname='documents:document_multiple_update_page_count',
-            data={'id_list': self.test_document.pk}
-        )
-
     def _request_document_clear_transformations_view(self):
         return self.post(
             viewname='documents:document_clear_transformations',
@@ -232,8 +248,11 @@ class DocumentViewTestMixin(object):
             data={'id_list': self.test_document.pk}
         )
 
-    def _request_empty_trash_view(self):
+    def _request_document_multiple_pages_reset_view(self):
-        return self.post(viewname='documents:trash_can_empty')
+        return self.post(
+            viewname='documents:document_multiple_pages_reset',
+            data={'id_list': self.test_document.pk}
+        )
 
     def _request_document_print_view(self):
         return self.get(
@@ -243,3 +262,6 @@ class DocumentViewTestMixin(object):
                 'page_group': PAGE_RANGE_ALL
             }
         )
+
+    def _request_empty_trash_view(self):
+        return self.post(viewname='documents:trash_can_empty')

mayan/apps/documents/tests/test_api.py

@@ -530,8 +530,7 @@ class DocumentPageAPIViewTestMixin(object):
         page = self.test_document.pages.first()
         return self.get(
             viewname='rest_api:documentpage-image', kwargs={
-                'pk': page.document.pk, 'version_pk': page.document_version.pk,
+                'pk': page.document.pk, 'page_pk': page.pk
-                'page_pk': page.pk
             }
         )
 
@@ -552,6 +551,33 @@ class DocumentPageAPIViewTestCase(
         self.assertEqual(response.status_code, status.HTTP_200_OK)
 
 
+class DocumentVersionPageAPIViewTestMixin(object):
+    def _request_document_version_page_image(self):
+        page = self.test_document_version.pages.first()
+        return self.get(
+            viewname='rest_api:documentversionpage-image', kwargs={
+                'pk': page.document.pk, 'version_pk': page.document_version.pk,
+                'page_pk': page.pk
+            }
+        )
+
+
+class DocumentVersionPageAPIViewTestCase(
+    DocumentVersionPageAPIViewTestMixin, DocumentTestMixin, BaseAPITestCase
+):
+    def test_document_version_page_api_image_view_no_access(self):
+        response = self._request_document_version_page_image()
+        self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+
+    def test_document_version_page_api_image_view_with_access(self):
+        self.grant_access(
+            obj=self.test_document, permission=permission_document_view
+        )
+
+        response = self._request_document_version_page_image()
+        self.assertEqual(response.status_code, status.HTTP_200_OK)
+
+
 class TrashedDocumentAPIViewTestMixin(object):
     def _request_test_document_api_trash_view(self):
         return self.delete(
@@ -575,13 +601,10 @@ class TrashedDocumentAPIViewTestMixin(object):
         )
 
     def _request_test_trashed_document_api_image_view(self):
-        latest_version = self.test_document.latest_version
-
         return self.get(
             viewname='rest_api:documentpage-image', kwargs={
-                'pk': latest_version.document.pk,
+                'pk': self.test_document.pk,
-                'version_pk': latest_version.pk,
+                'page_pk': self.test_document.pages.first().pk
-                'page_pk': latest_version.pages.first().pk
             }
         )
 

mayan/apps/documents/tests/test_document_page_views.py

@@ -9,10 +9,10 @@ from ..permissions import (
 from .base import GenericDocumentViewTestCase
 
 
-class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
+class DocumentPageDisableViewTestMixin(object):
-    def setUp(self):
+    def _disable_test_document_page(self):
-        super(DocumentPageDisableViewTestCase, self).setUp()
+        self.test_document_page.enabled = False
-        self.test_document_page = self.test_document.pages_all.first()
+        self.test_document_page.save()
 
     def _request_test_document_page_disable_view(self):
         return self.post(
@@ -21,6 +21,31 @@ class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
             }
         )
 
+    def _request_test_document_page_enable_view(self):
+        return self.post(
+            viewname='documents:document_page_enable', kwargs={
+                'pk': self.test_document_page.pk
+            }
+        )
+
+    def _request_test_document_page_multiple_disable_view(self):
+        return self.post(
+            viewname='documents:document_page_multiple_disable', data={
+                'id_list': self.test_document_page.pk
+            }
+        )
+
+    def _request_test_document_page_multiple_enable_view(self):
+        return self.post(
+            viewname='documents:document_page_multiple_enable', data={
+                'id_list': self.test_document_page.pk
+            }
+        )
+
+
+class DocumentPageDisableViewTestCase(
+    DocumentPageDisableViewTestMixin, GenericDocumentViewTestCase
+):
     def test_document_page_disable_view_no_permission(self):
         test_document_page_count = self.test_document.pages.count()
 
@@ -45,13 +70,6 @@ class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
             test_document_page_count, self.test_document.pages.count()
         )
 
-    def _request_test_document_page_multiple_disable_view(self):
-        return self.post(
-            viewname='documents:document_page_multiple_disable', data={
-                'id_list': self.test_document_page.pk
-            }
-        )
-
     def test_document_page_multiple_disable_view_no_permission(self):
         test_document_page_count = self.test_document.pages.count()
 
@@ -76,17 +94,6 @@ class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
             test_document_page_count, self.test_document.pages.count()
         )
 
-    def _disable_test_document_page(self):
-        self.test_document_page.enabled = False
-        self.test_document_page.save()
-
-    def _request_test_document_page_enable_view(self):
-        return self.post(
-            viewname='documents:document_page_enable', kwargs={
-                'pk': self.test_document_page.pk
-            }
-        )
-
     def test_document_page_enable_view_no_permission(self):
         self._disable_test_document_page()
 
@@ -114,13 +121,6 @@ class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
             test_document_page_count, self.test_document.pages.count()
         )
 
-    def _request_test_document_page_multiple_enable_view(self):
-        return self.post(
-            viewname='documents:document_page_multiple_enable', data={
-                'id_list': self.test_document_page.pk
-            }
-        )
-
     def test_document_page_multiple_enable_view_no_permission(self):
         self._disable_test_document_page()
         test_document_page_count = self.test_document.pages.count()
@@ -148,7 +148,7 @@ class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
         )
 
 
-class DocumentPageViewTestCase(GenericDocumentViewTestCase):
+class DocumentPageViewTestMixin(object):
     def _request_test_document_page_list_view(self):
         return self.get(
             viewname='documents:document_pages', kwargs={
@@ -156,6 +156,18 @@ class DocumentPageViewTestCase(GenericDocumentViewTestCase):
             }
         )
 
+    def _request_test_document_page_view(self, document_page):
+        return self.get(
+            viewname='documents:document_page_view', kwargs={
+                'pk': document_page.pk,
+            }
+        )
+
+
+class DocumentPageViewTestCase(
+    DocumentPageViewTestMixin, GenericDocumentViewTestCase
+):
+
     def test_document_page_list_view_no_permission(self):
         response = self._request_test_document_page_list_view()
         self.assertEqual(response.status_code, 404)
@@ -170,13 +182,6 @@ class DocumentPageViewTestCase(GenericDocumentViewTestCase):
             response=response, text=self.test_document.label, status_code=200
         )
 
-    def _request_test_document_page_view(self, document_page):
-        return self.get(
-            viewname='documents:document_page_view', kwargs={
-                'pk': document_page.pk,
-            }
-        )
-
     def test_document_page_view_no_permissions(self):
         response = self._request_test_document_page_view(
             document_page=self.test_document.pages.first()

mayan/apps/documents/tests/test_document_version_views.py

@@ -1,21 +1,19 @@
 from __future__ import unicode_literals
 
 from ..permissions import (
-    permission_document_version_revert, permission_document_version_view,
+    permission_document_tools, permission_document_version_revert,
+    permission_document_version_view,
 )
 
 from .base import GenericDocumentViewTestCase
 from .literals import TEST_VERSION_COMMENT
-from .mixins import DocumentVersionTestMixin
+from .mixins import DocumentVersionTestMixin, DocumentVersionViewTestMixin
 
 
-class DocumentVersionTestCase(DocumentVersionTestMixin, GenericDocumentViewTestCase):
+class DocumentVersionTestCase(
-    def _request_document_version_list_view(self):
+    DocumentVersionViewTestMixin, DocumentVersionTestMixin,
-        return self.get(
+    GenericDocumentViewTestCase
-            viewname='documents:document_version_list',
+):
-            kwargs={'pk': self.test_document.pk}
-        )
-
     def test_document_version_list_no_permission(self):
         self._upload_new_version()
 
@@ -33,12 +31,6 @@ class DocumentVersionTestCase(DocumentVersionTestMixin, GenericDocumentViewTestC
             response=response, text=TEST_VERSION_COMMENT, status_code=200
         )
 
-    def _request_document_version_revert_view(self, document_version):
-        return self.post(
-            viewname='documents:document_version_revert',
-            kwargs={'pk': document_version.pk}
-        )
-
     def test_document_version_revert_no_permission(self):
         first_version = self.test_document.latest_version
         self._upload_new_version()
@@ -64,3 +56,25 @@ class DocumentVersionTestCase(DocumentVersionTestMixin, GenericDocumentViewTestC
         self.assertEqual(response.status_code, 302)
 
         self.assertEqual(self.test_document.versions.count(), 1)
+
+    def test_document_version_page_count_update_view_no_permission(self):
+        self.test_document_version.pages.all().delete()
+
+        response = self._request_test_document_version_page_count_update_view()
+        self.assertEqual(response.status_code, 404)
+
+        self.assertEqual(self.test_document_version.pages.count(), 0)
+
+    def test_document_version_page_count_update_view_with_access(self):
+        page_count = self.test_document_version.pages.count()
+
+        self.test_document_version.pages.all().delete()
+
+        self.grant_access(
+            obj=self.test_document, permission=permission_document_tools
+        )
+
+        response = self._request_test_document_version_page_count_update_view()
+        self.assertEqual(response.status_code, 302)
+
+        self.assertEqual(self.test_document_version.pages.count(), page_count)

mayan/apps/documents/tests/test_document_views.py

@@ -292,46 +292,44 @@ class DocumentsViewsTestCase(
                 )
             )
 
-    def test_document_update_page_count_view_no_permission(self):
+    def test_document_pages_reset_view_no_permission(self):
         self.test_document.pages.all().delete()
-        self.assertEqual(self.test_document.pages.count(), 0)
 
-        response = self._request_document_update_page_count_view()
+        response = self._request_document_pages_reset_view()
         self.assertEqual(response.status_code, 404)
 
         self.assertEqual(self.test_document.pages.count(), 0)
 
-    def test_document_update_page_count_view_with_permission(self):
+    def test_document_pages_reset_view_with_access(self):
-        # TODO: Revise permission association
-
         page_count = self.test_document.pages.count()
         self.test_document.pages.all().delete()
-        self.assertEqual(self.test_document.pages.count(), 0)
 
-        self.grant_permission(permission=permission_document_tools)
+        self.grant_access(
+            obj=self.test_document, permission=permission_document_tools
+        )
 
-        response = self._request_document_update_page_count_view()
+        response = self._request_document_pages_reset_view()
         self.assertEqual(response.status_code, 302)
 
         self.assertEqual(self.test_document.pages.count(), page_count)
 
-    def test_document_multiple_update_page_count_view_no_permission(self):
+    def test_document_multiple_pages_reset_view_no_permission(self):
         self.test_document.pages.all().delete()
-        self.assertEqual(self.test_document.pages.count(), 0)
 
-        response = self._request_document_multiple_update_page_count_view()
+        response = self._request_document_multiple_pages_reset_view()
         self.assertEqual(response.status_code, 404)
 
         self.assertEqual(self.test_document.pages.count(), 0)
 
-    def test_document_multiple_update_page_count_view_with_permission(self):
+    def test_document_multiple_pages_reset_view_with_access(self):
         page_count = self.test_document.pages.count()
         self.test_document.pages.all().delete()
-        self.assertEqual(self.test_document.pages.count(), 0)
 
-        self.grant_permission(permission=permission_document_tools)
+        self.grant_access(
+            obj=self.test_document, permission=permission_document_tools
+        )
 
-        response = self._request_document_multiple_update_page_count_view()
+        response = self._request_document_multiple_pages_reset_view()
         self.assertEqual(response.status_code, 302)
 
         self.assertEqual(self.test_document.pages.count(), page_count)

mayan/apps/documents/tests/test_search.py

@@ -1,22 +1,30 @@
 from __future__ import unicode_literals
 
 from mayan.apps.common.tests.base import BaseTestCase
-from mayan.apps.documents.permissions import permission_document_view
+
-from mayan.apps.documents.search import document_search, document_page_search
+from ..permissions import permission_document_view
-from mayan.apps.documents.tests.mixins import DocumentTestMixin
+from ..search import document_search, document_page_search
+
+from .mixins import DocumentTestMixin
 
 
-class DocumentSearchTestCase(DocumentTestMixin, BaseTestCase):
+class DocumentSearchTestMixin(object):
     def _perform_document_page_search(self):
         return document_page_search.search(
-            query_string={'q': self.test_document.label}, user=self._test_case_user
+            query_string={'q': self.test_document.label},
+            user=self._test_case_user
         )
 
     def _perform_document_search(self):
         return document_search.search(
-            query_string={'q': self.test_document.label}, user=self._test_case_user
+            query_string={'q': self.test_document.label},
+            user=self._test_case_user
         )
 
+
+class DocumentSearchTestCase(
+    DocumentSearchTestMixin, DocumentTestMixin, BaseTestCase
+):
     def test_document_page_search_no_access(self):
         queryset = self._perform_document_page_search()
         self.assertFalse(self.test_document.pages.first() in queryset)

mayan/apps/documents/tests/test_trashed_document_views.py

@@ -9,7 +9,7 @@ from ..permissions import (
 from .base import GenericDocumentViewTestCase
 
 
-class TrashedDocumentTestCase(GenericDocumentViewTestCase):
+class TrashedDocumentTestMixin(object):
     def _request_document_restore_get_view(self):
         return self.get(
             viewname='documents:document_restore', kwargs={
@@ -17,6 +17,48 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
             }
         )
 
+    def _request_document_restore_post_view(self):
+        return self.post(
+            viewname='documents:document_restore', kwargs={
+                'pk': self.test_document.pk
+            }
+        )
+
+    def _request_document_trash_get_view(self):
+        return self.get(
+            viewname='documents:document_trash', kwargs={
+                'pk': self.test_document.pk
+            }
+        )
+
+    def _request_document_trash_post_view(self):
+        return self.post(
+            viewname='documents:document_trash', kwargs={
+                'pk': self.test_document.pk
+            }
+        )
+
+    def _request_trashed_document_delete_get_view(self):
+        return self.get(
+            viewname='documents:document_delete', kwargs={
+                'pk': self.test_document.pk
+            }
+        )
+
+    def _request_trashed_document_delete_post_view(self):
+        return self.post(
+            viewname='documents:document_delete', kwargs={
+                'pk': self.test_document.pk
+            }
+        )
+
+    def _request_trashed_document_list_view(self):
+        return self.get(viewname='documents:document_list_deleted')
+
+
+class TrashedDocumentTestCase(
+    TrashedDocumentTestMixin, GenericDocumentViewTestCase
+):
     def test_document_restore_get_view_no_permission(self):
         self.test_document.delete()
         self.assertEqual(Document.objects.count(), 0)
@@ -43,13 +85,6 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
 
         self.assertEqual(Document.objects.count(), document_count)
 
-    def _request_document_restore_post_view(self):
-        return self.post(
-            viewname='documents:document_restore', kwargs={
-                'pk': self.test_document.pk
-            }
-        )
-
     def test_document_restore_post_view_no_permission(self):
         self.test_document.delete()
         self.assertEqual(Document.objects.count(), 0)
@@ -74,13 +109,6 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
         self.assertEqual(DeletedDocument.objects.count(), 0)
         self.assertEqual(Document.objects.count(), 1)
 
-    def _request_document_trash_get_view(self):
-        return self.get(
-            viewname='documents:document_trash', kwargs={
-                'pk': self.test_document.pk
-            }
-        )
-
     def test_document_trash_get_view_no_permissions(self):
         document_count = Document.objects.count()
 
@@ -101,13 +129,6 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
 
         self.assertEqual(Document.objects.count(), document_count)
 
-    def _request_document_trash_post_view(self):
-        return self.post(
-            viewname='documents:document_trash', kwargs={
-                'pk': self.test_document.pk
-            }
-        )
-
     def test_document_trash_post_view_no_permissions(self):
         response = self._request_document_trash_post_view()
         self.assertEqual(response.status_code, 404)
@@ -126,13 +147,6 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
         self.assertEqual(DeletedDocument.objects.count(), 1)
         self.assertEqual(Document.objects.count(), 0)
 
-    def _request_document_delete_get_view(self):
-        return self.get(
-            viewname='documents:document_delete', kwargs={
-                'pk': self.test_document.pk
-            }
-        )
-
     def test_document_delete_get_view_no_permissions(self):
         self.test_document.delete()
         self.assertEqual(Document.objects.count(), 0)
@@ -140,7 +154,7 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
 
         trashed_document_count = DeletedDocument.objects.count()
 
-        response = self._request_document_delete_get_view()
+        response = self._request_trashed_document_delete_get_view()
         self.assertEqual(response.status_code, 404)
 
         self.assertEqual(
@@ -158,26 +172,19 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
 
         trashed_document_count = DeletedDocument.objects.count()
 
-        response = self._request_document_delete_get_view()
+        response = self._request_trashed_document_delete_get_view()
         self.assertEqual(response.status_code, 200)
 
         self.assertEqual(
             DeletedDocument.objects.count(), trashed_document_count
         )
 
-    def _request_document_delete_post_view(self):
-        return self.post(
-            viewname='documents:document_delete', kwargs={
-                'pk': self.test_document.pk
-            }
-        )
-
     def test_document_delete_post_view_no_permissions(self):
         self.test_document.delete()
         self.assertEqual(Document.objects.count(), 0)
         self.assertEqual(DeletedDocument.objects.count(), 1)
 
-        response = self._request_document_delete_post_view()
+        response = self._request_trashed_document_delete_post_view()
         self.assertEqual(response.status_code, 404)
 
         self.assertEqual(Document.objects.count(), 0)
@@ -192,19 +199,16 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
             obj=self.test_document, permission=permission_document_delete
         )
 
-        response = self._request_document_delete_post_view()
+        response = self._request_trashed_document_delete_post_view()
         self.assertEqual(response.status_code, 302)
 
         self.assertEqual(DeletedDocument.objects.count(), 0)
         self.assertEqual(Document.objects.count(), 0)
 
-    def _request_document_list_deleted_view(self):
-        return self.get(viewname='documents:document_list_deleted')
-
     def test_deleted_document_list_view_no_permissions(self):
         self.test_document.delete()
 
-        response = self._request_document_list_deleted_view()
+        response = self._request_trashed_document_list_view()
         self.assertNotContains(
             response=response, text=self.test_document.label, status_code=200
         )
@@ -216,7 +220,7 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
             obj=self.test_document, permission=permission_document_view
         )
 
-        response = self._request_document_list_deleted_view()
+        response = self._request_trashed_document_list_view()
         self.assertContains(
             response=response, text=self.test_document.label, status_code=200
         )

mayan/apps/documents/urls.py

@@ -4,21 +4,24 @@ from django.conf.urls import url
 
 from .api_views import (
     APITrashedDocumentListView, APIDeletedDocumentRestoreView,
-    APIDeletedDocumentView, APIDocumentDownloadView, APIDocumentView,
+    APIDeletedDocumentView, APIDocumentDownloadView, APIDocumentPageListView,
-    APIDocumentListView, APIDocumentVersionDownloadView,
+    APIDocumentView, APIDocumentListView, APIDocumentVersionDownloadView,
     APIDocumentPageImageView, APIDocumentPageView,
     APIDocumentTypeDocumentListView, APIDocumentTypeListView,
     APIDocumentTypeView, APIDocumentVersionsListView,
     APIDocumentVersionPageListView, APIDocumentVersionView,
-    APIRecentDocumentListView
+    APIRecentDocumentListView,
+    APIDocumentVersionPageView,
+    APIDocumentVersionPageImageView
 )
 from .views.document_views import (
     DocumentDocumentTypeEditView, DocumentDownloadFormView,
     DocumentDownloadView, DocumentDuplicatesListView, DocumentEditView,
     DocumentListView, DocumentPreviewView, DocumentPrint,
-    DocumentTransformationsClearView, DocumentTransformationsCloneView,
+    DocumentPagesResetView, DocumentTransformationsClearView,
-    DocumentUpdatePageCountView, DocumentView, DuplicatedDocumentListView,
+    DocumentTransformationsCloneView, DocumentView,
-    RecentAccessDocumentListView, RecentAddedDocumentListView
+    DuplicatedDocumentListView, RecentAccessDocumentListView,
+    RecentAddedDocumentListView
 )
 from .views.document_page_views import (
     DocumentPageDisable, DocumentPageEnable, DocumentPageListView,
@@ -30,7 +33,8 @@ from .views.document_page_views import (
 )
 from .views.document_version_views import (
     DocumentVersionDownloadFormView, DocumentVersionDownloadView,
-    DocumentVersionListView, DocumentVersionRevertView, DocumentVersionView,
+    DocumentVersionListView, DocumentVersionRevertView,
+    DocumentVersionUpdatePageCountView, DocumentVersionView,
 )
 from .views.document_type_views import (
     DocumentTypeCreateView, DocumentTypeDeleteView,
@@ -172,14 +176,14 @@ urlpatterns_documents = [
         name='document_print'
     ),
     url(
-        regex=r'^documents/(?P<pk>\d+)/reset_page_count/$',
+        regex=r'^documents/(?P<pk>\d+)/pages/reset/$',
-        view=DocumentUpdatePageCountView.as_view(),
+        view=DocumentPagesResetView.as_view(),
-        name='document_update_page_count'
+        name='document_pages_reset'
     ),
     url(
-        regex=r'^documents/multiple/reset_page_count/$',
+        regex=r'^documents/multiple/pages/reset/$',
-        view=DocumentUpdatePageCountView.as_view(),
+        view=DocumentPagesResetView.as_view(),
-        name='document_multiple_update_page_count'
+        name='document_multiple_pages_reset'
     ),
     url(
         regex=r'^documents/(?P<pk>\d+)/download/form/$',
@@ -305,6 +309,16 @@ urlpatterns_document_versions = [
         view=DocumentVersionDownloadView.as_view(),
         name='document_version_download'
     ),
+    url(
+        regex=r'^documents/versions/(?P<pk>\d+)/pages/update/$',
+        view=DocumentVersionUpdatePageCountView.as_view(),
+        name='document_version_page_count_update'
+    ),
+    url(
+        regex=r'^documents/versions/multiple/pages/update/$',
+        view=DocumentVersionUpdatePageCountView.as_view(),
+        name='document_version_multiple_page_count_update'
+    ),
     url(
         regex=r'^documents/versions/(?P<pk>\d+)/revert/$',
         view=DocumentVersionRevertView.as_view(),
@@ -405,6 +419,11 @@ api_urls = [
         view=APIDocumentVersionPageListView.as_view(),
         name='documentversion-page-list'
     ),
+    url(
+        regex=r'^documents/(?P<pk>[0-9]+)/pages/$',
+        view=APIDocumentPageListView.as_view(),
+        name='document-page-list'
+    ),
     url(
         regex=r'^documents/(?P<pk>[0-9]+)/versions/(?P<version_pk>[0-9]+)/download/$',
         view=APIDocumentVersionDownloadView.as_view(),
@@ -416,12 +435,20 @@ api_urls = [
     ),
     url(
         regex=r'^documents/(?P<pk>[0-9]+)/versions/(?P<version_pk>[0-9]+)/pages/(?P<page_pk>[0-9]+)$',
+        view=APIDocumentVersionPageView.as_view(), name='documentversionpage-detail'
+    ),
+    url(
+        regex=r'^documents/(?P<pk>[0-9]+)/pages/(?P<page_pk>[0-9]+)$',
         view=APIDocumentPageView.as_view(), name='documentpage-detail'
     ),
     url(
-        regex=r'^documents/(?P<pk>[0-9]+)/versions/(?P<version_pk>[0-9]+)/pages/(?P<page_pk>[0-9]+)/image/$',
+        regex=r'^documents/(?P<pk>[0-9]+)/pages/(?P<page_pk>[0-9]+)/image/$',
         view=APIDocumentPageImageView.as_view(), name='documentpage-image'
     ),
+    url(
+        regex=r'^documents/(?P<pk>[0-9]+)/versions/(?P<version_pk>[0-9]+)/pages/(?P<page_pk>[0-9]+)/image/$',
+        view=APIDocumentVersionPageImageView.as_view(), name='documentversionpage-image'
+    ),
     url(
         regex=r'^trashed_documents/$',
         view=APITrashedDocumentListView.as_view(), name='trasheddocument-list'

mayan/apps/documents/views/document_page_views.py

@@ -20,8 +20,8 @@ from mayan.apps.converter.literals import DEFAULT_ROTATION, DEFAULT_ZOOM_LEVEL
 
 from ..forms import DocumentPageForm
 from ..icons import icon_document_pages
-from ..links import link_document_update_page_count
+from ..links import link_document_pages_reset
-from ..models import Document, DocumentPage
+from ..models import Document, DocumentPage, DocumentVersionPage
 from ..permissions import permission_document_edit, permission_document_view
 from ..settings import (
     setting_rotation_step, setting_zoom_percent_step, setting_zoom_max_level,
@@ -50,13 +50,13 @@ class DocumentPageListView(ExternalObjectMixin, SingleObjectListView):
             'hide_object': True,
             'list_as_items': True,
             'no_results_icon': icon_document_pages,
-            'no_results_main_link': link_document_update_page_count.resolve(
+            'no_results_main_link': link_document_pages_reset.resolve(
                 request=self.request, resolved_object=self.external_object
             ),
             'no_results_text': _(
                 'This could mean that the document is of a format that is '
-                'not supported, that it is corrupted or that the upload '
+                'not supported, that it is corrupted, or that the upload '
-                'process was interrupted. Use the document page recalculation '
+                'process was interrupted. Use the document page reset '
                 'action to attempt to introspect the page count again.'
             ),
             'no_results_title': _('No document pages available'),

mayan/apps/documents/views/document_version_views.py

@@ -3,10 +3,11 @@ from __future__ import absolute_import, unicode_literals
 import logging
 
 from django.contrib import messages
-from django.utils.translation import ugettext_lazy as _
+from django.utils.translation import ugettext_lazy as _, ungettext
 
 from mayan.apps.common.generics import (
-    ConfirmView, SingleObjectDetailView, SingleObjectListView
+    ConfirmView, MultipleObjectConfirmActionView, SingleObjectDetailView,
+    SingleObjectListView
 )
 from mayan.apps.common.mixins import ExternalObjectMixin
 
@@ -14,9 +15,10 @@ from ..events import event_document_view
 from ..forms import DocumentVersionDownloadForm, DocumentVersionPreviewForm
 from ..models import Document, DocumentVersion
 from ..permissions import (
-    permission_document_download, permission_document_version_revert,
+    permission_document_download, permission_document_tools,
-    permission_document_version_view
+    permission_document_version_revert, permission_document_version_view
 )
+from ..tasks import task_update_page_count
 
 from .document_views import DocumentDownloadFormView, DocumentDownloadView
 
@@ -142,6 +144,45 @@ class DocumentVersionRevertView(ExternalObjectMixin, ConfirmView):
             )
 
 
+class DocumentVersionUpdatePageCountView(MultipleObjectConfirmActionView):
+    model = DocumentVersion
+    object_permission = permission_document_tools
+    success_message = _(
+        '%(count)d document version queued for page count recalculation'
+    )
+    success_message_plural = _(
+        '%(count)d documents version queued for page count recalculation'
+    )
+
+    def get_extra_context(self):
+        queryset = self.object_list
+
+        result = {
+            'title': ungettext(
+                singular='Recalculate the page count of the selected document version?',
+                plural='Recalculate the page count of the selected document versions?',
+                number=queryset.count()
+            )
+        }
+
+        if queryset.count() == 1:
+            result.update(
+                {
+                    'object': queryset.first(),
+                    'title': _(
+                        'Recalculate the page count of the document version: %s?'
+                    ) % queryset.first()
+                }
+            )
+
+        return result
+
+    def object_action(self, form, instance):
+        task_update_page_count.apply_async(
+            kwargs={'version_id': instance.pk}
+        )
+
+
 class DocumentVersionView(SingleObjectDetailView):
     form_class = DocumentVersionPreviewForm
     model = DocumentVersion

mayan/apps/documents/views/document_views.py

@@ -44,14 +44,14 @@ from ..permissions import (
 from ..settings import (
     setting_print_width, setting_print_height, setting_recent_added_count
 )
-from ..tasks import task_update_page_count
+from ..tasks import task_document_pages_reset
 from ..utils import parse_range
 
 __all__ = (
     'DocumentListView', 'DocumentDocumentTypeEditView',
     'DocumentDuplicatesListView', 'DocumentEditView', 'DocumentPreviewView',
     'DocumentView', 'DocumentDownloadFormView', 'DocumentDownloadView',
-    'DocumentUpdatePageCountView', 'DocumentTransformationsClearView',
+    'DocumentPagesResetView', 'DocumentTransformationsClearView',
     'DocumentTransformationsCloneView', 'DocumentPrint',
     'DuplicatedDocumentListView', 'RecentAccessDocumentListView',
     'RecentAddedDocumentListView'
@@ -418,6 +418,52 @@ class DocumentPreviewView(SingleObjectDetailView):
         }
 
 
+class DocumentPagesResetView(MultipleObjectConfirmActionView):
+    model = Document
+    object_permission = permission_document_tools
+    success_message = _('%(count)d document queued for pages reset')
+    success_message_plural = _('%(count)d documents queued for pages reset')
+
+    def get_extra_context(self):
+        queryset = self.object_list
+
+        result = {
+            'title': ungettext(
+                singular='Reset the pages of the selected document?',
+                plural='Reset the pages of the selected documents?',
+                number=queryset.count()
+            )
+        }
+
+        if queryset.count() == 1:
+            result.update(
+                {
+                    'object': queryset.first(),
+                    'title': _(
+                        'Reset the pages of the document: %s?'
+                    ) % queryset.first()
+                }
+            )
+
+        return result
+
+    def object_action(self, form, instance):
+        latest_version = instance.latest_version
+        if latest_version:
+            task_document_pages_reset.apply_async(
+                kwargs={'document_id': instance.pk}
+            )
+        else:
+            messages.error(
+                self.request, _(
+                    'Document "%(document)s" is empty. Upload at least one '
+                    'document version before attempting to reset the pages. '
+                ) % {
+                    'document': instance,
+                }
+            )
+
+
 class DocumentView(SingleObjectDetailView):
     form_class = DocumentPropertiesForm
     model = Document
@@ -436,57 +482,6 @@ class DocumentView(SingleObjectDetailView):
         }
 
 
-class DocumentUpdatePageCountView(MultipleObjectConfirmActionView):
-    model = Document
-    object_permission = permission_document_tools
-    success_message = _(
-        '%(count)d document queued for page count recalculation'
-    )
-    success_message_plural = _(
-        '%(count)d documents queued for page count recalculation'
-    )
-
-    def get_extra_context(self):
-        queryset = self.object_list
-
-        result = {
-            'title': ungettext(
-                singular='Recalculate the page count of the selected document?',
-                plural='Recalculate the page count of the selected documents?',
-                number=queryset.count()
-            )
-        }
-
-        if queryset.count() == 1:
-            result.update(
-                {
-                    'object': queryset.first(),
-                    'title': _(
-                        'Recalculate the page count of the document: %s?'
-                    ) % queryset.first()
-                }
-            )
-
-        return result
-
-    def object_action(self, form, instance):
-        latest_version = instance.latest_version
-        if latest_version:
-            task_update_page_count.apply_async(
-                kwargs={'version_id': latest_version.pk}
-            )
-        else:
-            messages.error(
-                self.request, _(
-                    'Document "%(document)s" is empty. Upload at least one '
-                    'document version before attempting to detect the '
-                    'page count.'
-                ) % {
-                    'document': instance,
-                }
-            )
-
-
 class DocumentTransformationsClearView(MultipleObjectConfirmActionView):
     model = Document
     object_permission = permission_transformation_delete

mayan/apps/dynamic_search/classes.py

@@ -184,7 +184,14 @@ class SearchModel(object):
             query_string=query_string, global_and_search=global_and_search
         )
 
-        queryset = self.get_queryset().filter(search_query.query).distinct()
+        try:
+            queryset = self.get_queryset().filter(search_query.query).distinct()
+        except Exception:
+            logger.error(
+                'Error filtering model %s with queryset: %s', self.model,
+                search_query.query
+            )
+            raise
 
         if self.permission:
             queryset = AccessControlList.objects.restrict_queryset(

mayan/apps/file_caching/migrations/0003_auto_20191008_1510.py

@@ -0,0 +1,20 @@
+# -*- coding: utf-8 -*-
+# Generated by Django 1.11.22 on 2019-10-08 15:10
+from __future__ import unicode_literals
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('file_caching', '0002_auto_20190729_0236'),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name='cache',
+            name='name',
+            field=models.CharField(db_index=True, help_text='Internal name of the cache.', max_length=128, unique=True, verbose_name='Name'),
+        ),
+    ]

mayan/apps/file_metadata/apps.py

@@ -12,7 +12,9 @@ from mayan.apps.common.menus import (
     menu_tools
 )
 from mayan.apps.document_indexing.handlers import handler_index_document
-from mayan.apps.documents.search import document_page_search, document_search
+from mayan.apps.documents.search import (
+    document_page_search, document_search, document_version_page_search
+)
 from mayan.apps.documents.signals import post_version_upload
 from mayan.apps.events.classes import ModelEventType
 from mayan.apps.navigation.classes import SourceColumn
@@ -156,11 +158,19 @@ class FileMetadataApp(MayanAppConfig):
             label=_('File metadata value')
         )
 
-        document_page_search.add_model_field(
+        #document_page_search.add_model_field(
+        #    field='document__document_version__file_metadata_drivers__entries__key',
+        #    label=_('File metadata key')
+        #)
+        #document_page_search.add_model_field(
+        #    field='document__document_version__file_metadata_drivers__entries__value',
+        #    label=_('File metadata value')
+        #)
+        document_version_page_search.add_model_field(
             field='document_version__file_metadata_drivers__entries__key',
             label=_('File metadata key')
         )
-        document_page_search.add_model_field(
+        document_version_page_search.add_model_field(
             field='document_version__file_metadata_drivers__entries__value',
             label=_('File metadata value')
         )

mayan/apps/metadata/apps.py

@@ -16,7 +16,9 @@ from mayan.apps.common.menus import (
     menu_facet, menu_list_facet, menu_multi_item, menu_object, menu_secondary,
     menu_setup
 )
-from mayan.apps.documents.search import document_page_search, document_search
+from mayan.apps.documents.search import (
+    document_page_search, document_search, document_version_page_search
+)
 from mayan.apps.documents.signals import post_document_type_change
 from mayan.apps.events.classes import ModelEventType
 from mayan.apps.events.links import (
@@ -76,7 +78,7 @@ class MetadataApp(MayanAppConfig):
             app_label='documents', model_name='Document'
         )
         DocumentPageResult = apps.get_model(
-            app_label='documents', model_name='DocumentPageResult'
+            app_label='documents', model_name='DocumentVersionPageResult'
         )
 
         DocumentType = apps.get_model(
@@ -188,10 +190,18 @@ class MetadataApp(MayanAppConfig):
         )
 
         document_page_search.add_model_field(
-            field='document_version__document__metadata__metadata_type__name',
+            field='document__metadata__metadata_type__name',
             label=_('Metadata type')
         )
         document_page_search.add_model_field(
+            field='document__metadata__value',
+            label=_('Metadata value')
+        )
+        document_version_page_search.add_model_field(
+            field='document_version__document__metadata__metadata_type__name',
+            label=_('Metadata type')
+        )
+        document_version_page_search.add_model_field(
             field='document_version__document__metadata__value',
             label=_('Metadata value')
         )

mayan/apps/ocr/admin.py

@@ -3,13 +3,15 @@ from __future__ import unicode_literals
 from django.contrib import admin
 
 from .models import (
-    DocumentPageOCRContent, DocumentTypeSettings, DocumentVersionOCRError
+    DocumentTypeSettings, DocumentVersionPageOCRContent,
+    DocumentVersionOCRError
 )
 
 
-@admin.register(DocumentPageOCRContent)
+@admin.register(DocumentVersionPageOCRContent)
-class DocumentPageOCRContentAdmin(admin.ModelAdmin):
+class DocumentVersionPageOCRContentAdmin(admin.ModelAdmin):
-    list_display = ('document_page',)
+    pass
+    #list_display = ('document_page',)
 
 
 @admin.register(DocumentTypeSettings)

mayan/apps/ocr/api_views.py

@@ -8,7 +8,7 @@ from rest_framework.response import Response
 from mayan.apps.documents.models import Document, DocumentVersion
 from mayan.apps.rest_api.permissions import MayanPermission
 
-from .models import DocumentPageOCRContent
+from .models import DocumentVersionPageOCRContent
 from .permissions import permission_ocr_content_view, permission_ocr_document
 from .serializers import DocumentPageOCRContentSerializer
 
@@ -90,8 +90,8 @@ class APIDocumentPageOCRContentView(generics.RetrieveAPIView):
 
         try:
             ocr_content = instance.ocr_content
-        except DocumentPageOCRContent.DoesNotExist:
+        except DocumentVersionPageOCRContent.DoesNotExist:
-            ocr_content = DocumentPageOCRContent.objects.none()
+            ocr_content = DocumentVersionPageOCRContent.objects.none()
 
         serializer = self.get_serializer(ocr_content)
         return Response(serializer.data)

mayan/apps/ocr/apps.py

@@ -12,7 +12,9 @@ from mayan.apps.common.classes import ModelField
 from mayan.apps.common.menus import (
     menu_facet, menu_list_facet, menu_multi_item, menu_secondary, menu_tools
 )
-from mayan.apps.documents.search import document_search, document_page_search
+from mayan.apps.documents.search import (
+    document_search, document_page_search, document_version_page_search
+)
 from mayan.apps.documents.signals import post_version_upload
 from mayan.apps.events.classes import ModelEventType
 from mayan.apps.navigation.classes import SourceColumn
@@ -32,17 +34,19 @@ from .links import (
     link_document_ocr_content_delete_multiple, link_document_ocr_download,
     link_document_ocr_errors_list, link_document_submit,
     link_document_submit_multiple, link_document_type_ocr_settings,
-    link_document_type_submit, link_entry_list
+    link_document_type_submit, link_document_version_page_ocr_content,
+    link_entry_list
 )
 from .methods import (
-    method_document_ocr_submit, method_document_version_ocr_submit
+    method_document_ocr_submit, method_document_page_get_ocr_content,
+    method_document_version_ocr_submit
 )
 from .permissions import (
     permission_document_type_ocr_setup, permission_ocr_document,
     permission_ocr_content_view
 )
 from .signals import post_document_version_ocr
-from .utils import get_document_ocr_content
+from .utils import get_document_ocr_content, get_document_version_ocr_content
 
 logger = logging.getLogger(__name__)
 
@@ -73,6 +77,9 @@ class OCRApp(MayanAppConfig):
         DocumentVersion = apps.get_model(
             app_label='documents', model_name='DocumentVersion'
         )
+        DocumentVersionPage = apps.get_model(
+            app_label='documents', model_name='DocumentVersionPage'
+        )
 
         DocumentVersionOCRError = self.get_model(
             model_name='DocumentVersionOCRError'
@@ -81,8 +88,11 @@ class OCRApp(MayanAppConfig):
         Document.add_to_class(
             name='submit_for_ocr', value=method_document_ocr_submit
         )
+        DocumentPage.add_to_class(
+            name='get_ocr_content', value=method_document_page_get_ocr_content
+        )
         DocumentVersion.add_to_class(
-            name='ocr_content', value=get_document_ocr_content
+            name='ocr_content', value=get_document_version_ocr_content
         )
         DocumentVersion.add_to_class(
             name='submit_for_ocr', value=method_document_version_ocr_submit
@@ -97,7 +107,7 @@ class OCRApp(MayanAppConfig):
         )
 
         ModelField(
-            model=Document, name='versions__version_pages__ocr_content__content'
+            model=Document, name='versions__pages__ocr_content__content'
         )
 
         ModelPermission.register(
@@ -128,12 +138,14 @@ class OCRApp(MayanAppConfig):
         )
 
         document_search.add_model_field(
-            field='versions__version_pages__ocr_content__content', label=_('OCR')
+            field='versions__pages__ocr_content__content', label=_('OCR')
         )
-
+        document_version_page_search.add_model_field(
-        document_page_search.add_model_field(
             field='ocr_content__content', label=_('OCR')
         )
+        #document_page_search.add_model_field(
+        #    field='ocr_content__content', label=_('OCR')
+        #)
 
         menu_facet.bind_links(
             links=(link_document_ocr_content,), sources=(Document,)
@@ -141,6 +153,10 @@ class OCRApp(MayanAppConfig):
         menu_list_facet.bind_links(
             links=(link_document_page_ocr_content,), sources=(DocumentPage,)
         )
+        menu_list_facet.bind_links(
+            links=(link_document_version_page_ocr_content,),
+            sources=(DocumentVersionPage,)
+        )
         menu_list_facet.bind_links(
             links=(link_document_type_ocr_settings,), sources=(DocumentType,)
         )

mayan/apps/ocr/forms.py

@@ -8,7 +8,7 @@ from django.utils.translation import ugettext_lazy as _, ugettext
 
 from mayan.apps.common.widgets import TextAreaDiv
 
-from .models import DocumentPageOCRContent
+from .models import DocumentVersionPageOCRContent
 
 
 class DocumentPageOCRContentForm(forms.Form):
@@ -28,15 +28,26 @@ class DocumentPageOCRContentForm(forms.Form):
         content = ''
         self.fields['contents'].initial = ''
 
-        try:
+        content = conditional_escape(
-            page_content = page.ocr_content.content
+            force_text(self.get_instance_ocr_content(instance=page))
-        except DocumentPageOCRContent.DoesNotExist:
+        )
-            pass
-        else:
-            content = conditional_escape(force_text(page_content))
 
         self.fields['contents'].initial = mark_safe(content)
 
+    def get_instance_ocr_content(self, instance):
+        try:
+            return instance.content_object.ocr_content.content
+        except DocumentVersionPageOCRContent.DoesNotExist:
+            return ''
+
+
+class DocumentVersionPageOCRContentForm(DocumentPageOCRContentForm):
+    def get_instance_ocr_content(self, instance):
+        try:
+            return instance.ocr_content.content
+        except (AttributeError, DocumentVersionPageOCRContent.DoesNotExist):
+            return ''
+
 
 class DocumentOCRContentForm(forms.Form):
     """
@@ -54,19 +65,15 @@ class DocumentOCRContentForm(forms.Form):
     )
 
     def __init__(self, *args, **kwargs):
-        self.document = kwargs.pop('instance', None)
+        document = kwargs.pop('instance', None)
         super(DocumentOCRContentForm, self).__init__(*args, **kwargs)
         content = []
         self.fields['contents'].initial = ''
-        try:
-            document_pages = self.document.pages.all()
-        except AttributeError:
-            document_pages = []
 
-        for page in document_pages:
+        for document_page in document.pages.all():
             try:
-                page_content = page.ocr_content.content
+                page_content = document_page.content_object.ocr_content.content
-            except DocumentPageOCRContent.DoesNotExist:
+            except (AttributeError, DocumentVersionPageOCRContent.DoesNotExist):
                 pass
             else:
                 content.append(conditional_escape(force_text(page_content)))
@@ -74,7 +81,7 @@ class DocumentOCRContentForm(forms.Form):
                     '\n\n\n<hr/><div class="document-page-content-divider">- %s -</div><hr/>\n\n\n' % (
                         ugettext(
                             'Page %(page_number)d'
-                        ) % {'page_number': page.page_number}
+                        ) % {'page_number': document_page.page_number}
                     )
                 )
 

mayan/apps/ocr/icons.py

@@ -19,7 +19,7 @@ icon_document_ocr_errors_list = Icon(
 icon_document_type_ocr_settings = Icon(
     driver_name='fontawesome', symbol='font'
 )
-icon_document_type_submit = Icon(driver_name='fontawesome', symbol='font')
-icon_entry_list = Icon(driver_name='fontawesome', symbol='font')
-
 icon_document_submit = icon_document_multiple_submit
+icon_document_type_submit = Icon(driver_name='fontawesome', symbol='font')
+icon_document_version_page_ocr_content = Icon(driver_name='fontawesome', symbol='font')
+icon_entry_list = Icon(driver_name='fontawesome', symbol='font')

mayan/apps/ocr/links.py

@@ -58,10 +58,11 @@ link_document_type_submit = Link(
     permissions=(permission_ocr_document,), text=_('OCR documents per type'),
     view='ocr:document_type_submit'
 )
-link_entry_list = Link(
+link_document_version_page_ocr_content = Link(
-    icon_class_path='mayan.apps.ocr.icons.icon_entry_list',
+    args='resolved_object.id',
-    permissions=(permission_ocr_document,), text=_('OCR errors'),
+    icon_class_path='mayan.apps.ocr.icons.icon_document_version_page_ocr_content',
-    view='ocr:entry_list'
+    permissions=(permission_ocr_content_view,), text=_('OCR'),
+    view='ocr:document_version_page_ocr_content',
 )
 link_document_ocr_errors_list = Link(
     args='resolved_object.id',
@@ -75,3 +76,8 @@ link_document_ocr_download = Link(
     permissions=(permission_ocr_content_view,), text=_('Download OCR text'),
     view='ocr:document_ocr_download'
 )
+link_entry_list = Link(
+    icon_class_path='mayan.apps.ocr.icons.icon_entry_list',
+    permissions=(permission_ocr_document,), text=_('OCR errors'),
+    view='ocr:entry_list'
+)

mayan/apps/ocr/managers.py

@@ -9,7 +9,9 @@ from django.conf import settings
 from django.db import models, transaction
 
 from mayan.apps.documents.literals import DOCUMENT_IMAGE_TASK_TIMEOUT
-from mayan.apps.documents.tasks import task_generate_document_page_image
+from mayan.apps.documents.tasks import (
+    task_generate_document_version_page_image
+)
 
 from .events import (
     event_ocr_document_content_deleted, event_ocr_document_version_finish
@@ -20,47 +22,53 @@ from .signals import post_document_version_ocr
 logger = logging.getLogger(__name__)
 
 
-class DocumentPageOCRContentManager(models.Manager):
+class DocumentVesionPageOCRContentManager(models.Manager):
     def delete_content_for(self, document, user=None):
         with transaction.atomic():
             for document_page in document.pages.all():
-                self.filter(document_page=document_page).delete()
+                self.filter(
+                    document_version_page=document_page.content_object
+                ).delete()
 
             event_ocr_document_content_deleted.commit(
                 actor=user, target=document
             )
 
-    def process_document_page(self, document_page):
+    def process_document_version_page(self, document_version_page):
         logger.info(
             'Processing page: %d of document version: %s',
-            document_page.page_number, document_page.document_version
+            document_version_page.page_number,
+            document_version_page.document_version
         )
 
-        DocumentPageOCRContent = apps.get_model(
+        DocumentVersionPageOCRContent = apps.get_model(
-            app_label='ocr', model_name='DocumentPageOCRContent'
+            app_label='ocr', model_name='DocumentVersionPageOCRContent'
         )
 
-        task = task_generate_document_page_image.apply_async(
+        task = task_generate_document_version_page_image.apply_async(
             kwargs=dict(
-                document_page_id=document_page.pk
+                document_version_page_id=document_version_page.pk
             )
         )
 
-        cache_filename = task.get(timeout=DOCUMENT_IMAGE_TASK_TIMEOUT, disable_sync_subtasks=False)
+        cache_filename = task.get(
+            timeout=DOCUMENT_IMAGE_TASK_TIMEOUT, disable_sync_subtasks=False
+        )
 
-        with document_page.cache_partition.get_file(filename=cache_filename).open() as file_object:
+        with document_version_page.cache_partition.get_file(filename=cache_filename).open() as file_object:
-            document_page_content, created = DocumentPageOCRContent.objects.get_or_create(
+            document_version_page_content, created = DocumentVersionPageOCRContent.objects.get_or_create(
-                document_page=document_page
+                document_version_page=document_version_page
             )
-            document_page_content.content = ocr_backend.execute(
+            document_version_page_content.content = ocr_backend.execute(
                 file_object=file_object,
-                language=document_page.document.language
+                language=document_version_page.document.language
             )
-            document_page_content.save()
+            document_version_page_content.save()
 
         logger.info(
             'Finished processing page: %d of document version: %s',
-            document_page.page_number, document_page.document_version
+            document_version_page.page_number,
+            document_version_page.document_version
         )
 
     def process_document_version(self, document_version):
@@ -68,8 +76,10 @@ class DocumentPageOCRContentManager(models.Manager):
         logger.debug('document version: %d', document_version.pk)
 
         try:
-            for document_page in document_version.pages.all():
+            for document_version_page in document_version.pages.all():
-                self.process_document_page(document_page=document_page)
+                self.process_document_version_page(
+                    document_version_page=document_version_page
+                )
         except Exception as exception:
             logger.error(
                 'OCR error for document version: %d; %s', document_version.pk,

mayan/apps/ocr/methods.py

@@ -2,6 +2,7 @@ from __future__ import unicode_literals
 
 from datetime import timedelta
 
+from django.apps import apps
 from django.utils.timezone import now
 
 from mayan.apps.common.settings import settings_db_sync_task_delay
@@ -17,6 +18,17 @@ def method_document_ocr_submit(self):
         latest_version.submit_for_ocr()
 
 
+def method_document_page_get_ocr_content(self):
+    DocumentVersionPageOCRContent = apps.get_model(
+        app_label='ocr', model_name='DocumentVersionPageOCRContent'
+    )
+
+    try:
+        return self.content_object.ocr_content.content
+    except (AttributeError, DocumentVersionPageOCRContent.DoesNotExist):
+        return None
+
+
 def method_document_version_ocr_submit(self):
     event_ocr_document_version_submit.commit(
         action_object=self.document, target=self

mayan/apps/ocr/migrations/0009_rename_page_content.py

@@ -0,0 +1,42 @@
+from __future__ import unicode_literals
+
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ('ocr', '0008_auto_20180917_0646'),
+        ('documents', '0052_rename_document_page'),
+    ]
+
+    operations = [
+        migrations.RenameModel(
+            'DocumentPageOCRContent', 'DocumentVersionPageOCRContent'
+        ),
+        migrations.AlterField(
+            model_name='documentversionpageocrcontent',
+            name='document_page',
+            field=models.OneToOneField(
+                on_delete=django.db.models.deletion.CASCADE,
+                #name='document_version_page',
+                related_name='ocr_content',
+                to='documents.DocumentVersionPage',
+                verbose_name='Document version page'
+            ),
+        ),
+        migrations.RenameField(
+            model_name='documentversionpageocrcontent',
+            old_name='document_page',
+            new_name='document_version_page',
+        ),
+        migrations.AlterModelOptions(
+            name='documentversionpageocrcontent',
+            options={
+                'verbose_name': 'Document version page OCR content',
+                'verbose_name_plural': 'Document version pages OCR contents'
+            },
+        ),
+    ]
+
+

mayan/apps/ocr/models.py

@@ -4,10 +4,12 @@ from django.db import models
 from django.utils.encoding import force_text, python_2_unicode_compatible
 from django.utils.translation import ugettext_lazy as _
 
-from mayan.apps.documents.models import DocumentPage, DocumentType, DocumentVersion
+from mayan.apps.documents.models import (
+    DocumentPage, DocumentType, DocumentVersion, DocumentVersionPage
+)
 
 from .managers import (
-    DocumentPageOCRContentManager, DocumentTypeSettingsManager
+    DocumentVesionPageOCRContentManager, DocumentTypeSettingsManager
 )
 
 
@@ -36,13 +38,13 @@ class DocumentTypeSettings(models.Model):
 
 
 @python_2_unicode_compatible
-class DocumentPageOCRContent(models.Model):
+class DocumentVersionPageOCRContent(models.Model):
     """
     This model stores the OCR results for a document page.
     """
-    document_page = models.OneToOneField(
+    document_version_page = models.OneToOneField(
         on_delete=models.CASCADE, related_name='ocr_content',
-        to=DocumentPage, verbose_name=_('Document page')
+        to=DocumentVersionPage, verbose_name=_('Document version page')
     )
     content = models.TextField(
         blank=True, help_text=_(
@@ -50,11 +52,11 @@ class DocumentPageOCRContent(models.Model):
         ), verbose_name=_('Content')
     )
 
-    objects = DocumentPageOCRContentManager()
+    objects = DocumentVesionPageOCRContentManager()
 
     class Meta:
-        verbose_name = _('Document page OCR content')
+        verbose_name = _('Document version page OCR content')
-        verbose_name_plural = _('Document pages OCR contents')
+        verbose_name_plural = _('Document version pages OCR contents')
 
     def __str__(self):
         return force_text(self.document_page)

mayan/apps/ocr/serializers.py

@@ -2,10 +2,10 @@ from __future__ import unicode_literals
 
 from rest_framework import serializers
 
-from .models import DocumentPageOCRContent
+from .models import DocumentVersionPageOCRContent
 
 
 class DocumentPageOCRContentSerializer(serializers.ModelSerializer):
     class Meta:
         fields = ('content',)
-        model = DocumentPageOCRContent
+        model = DocumentVersionPageOCRContent

mayan/apps/ocr/tasks.py

@@ -19,8 +19,8 @@ def task_do_ocr(self, document_version_pk):
     DocumentVersion = apps.get_model(
         app_label='documents', model_name='DocumentVersion'
     )
-    DocumentPageOCRContent = apps.get_model(
+    DocumentVersionPageOCRContent = apps.get_model(
-        app_label='ocr', model_name='DocumentPageOCRContent'
+        app_label='ocr', model_name='DocumentVersionPageOCRContent'
     )
 
     lock_id = 'task_do_ocr_doc_version-%d' % document_version_pk
@@ -39,7 +39,7 @@ def task_do_ocr(self, document_version_pk):
                 'Starting document OCR for document version: %s',
                 document_version
             )
-            DocumentPageOCRContent.objects.process_document_version(
+            DocumentVersionPageOCRContent.objects.process_document_version(
                 document_version=document_version
             )
         except OperationalError as exception:

mayan/apps/ocr/tests/test_api.py

@@ -12,13 +12,34 @@ from ..permissions import (
 from .literals import TEST_DOCUMENT_CONTENT
 
 
-class OCRAPITestCase(DocumentTestMixin, BaseAPITestCase):
+class OCRAPIViewTestMixin(object):
     def _request_document_ocr_submit_view(self):
         return self.post(
             viewname='rest_api:document-ocr-submit-view',
             kwargs={'pk': self.test_document.pk}
         )
 
+    def _request_document_version_ocr_submit_view(self):
+        return self.post(
+            viewname='rest_api:document-version-ocr-submit-view', kwargs={
+                'document_pk': self.test_document.pk,
+                'version_pk': self.test_document.latest_version.pk
+            }
+        )
+
+    def _request_document_version_page_content_view(self):
+        return self.get(
+            viewname='rest_api:document-page-ocr-content-view', kwargs={
+                'document_pk': self.test_document.pk,
+                'version_pk': self.test_document.latest_version.pk,
+                'page_pk': self.test_document.latest_version.pages.first().pk,
+            }
+        )
+
+
+class OCRAPIViewTestCase(
+    OCRAPIViewTestMixin, DocumentTestMixin, BaseAPITestCase
+):
     def test_submit_document_no_access(self):
         response = self._request_document_ocr_submit_view()
         self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
@@ -35,15 +56,9 @@ class OCRAPITestCase(DocumentTestMixin, BaseAPITestCase):
         self.assertEqual(response.status_code, status.HTTP_202_ACCEPTED)
 
         self.assertTrue(
-            hasattr(self.test_document.pages.first(), 'ocr_content')
+            hasattr(
-        )
+                self.test_document.pages.first().content_object, 'ocr_content'
-
+            )
-    def _request_document_version_ocr_submit_view(self):
-        return self.post(
-            viewname='rest_api:document-version-ocr-submit-view', kwargs={
-                'document_pk': self.test_document.pk,
-                'version_pk': self.test_document.latest_version.pk
-            }
         )
 
     def test_submit_document_version_no_access(self):
@@ -62,20 +77,11 @@ class OCRAPITestCase(DocumentTestMixin, BaseAPITestCase):
         self.assertEqual(response.status_code, status.HTTP_202_ACCEPTED)
 
         self.assertTrue(
-            hasattr(self.test_document.pages.first(), 'ocr_content')
+            hasattr(self.test_document_version.pages.first(), 'ocr_content')
-        )
-
-    def _request_document_page_content_view(self):
-        return self.get(
-            viewname='rest_api:document-page-ocr-content-view', kwargs={
-                'document_pk': self.test_document.pk,
-                'version_pk': self.test_document.latest_version.pk,
-                'page_pk': self.test_document.latest_version.pages.first().pk,
-            }
         )
 
     def test_get_document_version_page_content_no_access(self):
-        response = self._request_document_page_content_view()
+        response = self._request_document_version_page_content_view()
         self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
 
     def test_get_document_version_page_content_with_access(self):
@@ -83,7 +89,7 @@ class OCRAPITestCase(DocumentTestMixin, BaseAPITestCase):
         self.grant_access(
             permission=permission_ocr_content_view, obj=self.test_document
         )
-        response = self._request_document_page_content_view()
+        response = self._request_document_version_page_content_view()
         self.assertEqual(response.status_code, status.HTTP_200_OK)
         self.assertTrue(
             TEST_DOCUMENT_CONTENT in response.data['content']

mayan/apps/ocr/tests/test_events.py

@@ -8,13 +8,13 @@ from ..events import (
     event_ocr_document_content_deleted, event_ocr_document_version_submit,
     event_ocr_document_version_finish
 )
-from ..models import DocumentPageOCRContent
+from ..models import DocumentVersionPageOCRContent
 
 
 class OCREventsTestCase(GenericDocumentTestCase):
     def test_document_content_deleted_event(self):
         Action.objects.all().delete()
-        DocumentPageOCRContent.objects.delete_content_for(
+        DocumentVersionPageOCRContent.objects.delete_content_for(
             document=self.test_document
         )
 

mayan/apps/ocr/tests/test_models.py

@@ -19,7 +19,7 @@ class DocumentOCRTestCase(DocumentTestMixin, BaseTestCase):
     _skip_file_descriptor_test = True
 
     def test_ocr_language_backends_end(self):
-        content = self.test_document.pages.first().ocr_content.content
+        content = self.test_document.pages.first().content_object.ocr_content.content
         self.assertTrue(TEST_DOCUMENT_CONTENT in content)
 
 
@@ -40,7 +40,7 @@ class GermanOCRSupportTestCase(DocumentTestMixin, BaseTestCase):
             )
 
     def test_ocr_language_backends_end(self):
-        content = self.test_document.pages.first().ocr_content.content
+        content = self.test_document.pages.first().content_object.ocr_content.content
 
         self.assertTrue(
             TEST_DOCUMENT_CONTENT_DEU_1 in content

mayan/apps/ocr/tests/test_views.py

@@ -2,7 +2,7 @@ from __future__ import unicode_literals
 
 from mayan.apps.documents.tests.base import GenericDocumentViewTestCase
 
-from ..models import DocumentPageOCRContent
+from ..models import DocumentVersionPageOCRContent
 from ..permissions import (
     permission_ocr_content_view, permission_ocr_document,
     permission_document_type_ocr_setup
@@ -27,10 +27,10 @@ class OCRViewTestMixin(object):
             }
         )
 
-    def _request_document_page_content_view(self):
+    def _request_document_version_page_content_view(self):
         return self.get(
-            viewname='ocr:document_page_ocr_content', kwargs={
+            viewname='ocr:document_version_page_ocr_content', kwargs={
-                'pk': self.test_document.pages.first().pk
+                'pk': self.test_document_version.pages.first().pk
             }
         )
 
@@ -86,8 +86,8 @@ class OCRViewsTestCase(OCRViewTestMixin, GenericDocumentViewTestCase):
         self.assertEqual(response.status_code, 404)
 
         self.assertTrue(
-            DocumentPageOCRContent.objects.filter(
+            DocumentVersionPageOCRContent.objects.filter(
-                document_page=self.test_document.pages.first()
+                document_version_page=self.test_document.pages.first().content_object
             ).exists()
         )
 
@@ -101,28 +101,11 @@ class OCRViewsTestCase(OCRViewTestMixin, GenericDocumentViewTestCase):
         self.assertEqual(response.status_code, 302)
 
         self.assertFalse(
-            DocumentPageOCRContent.objects.filter(
+            DocumentVersionPageOCRContent.objects.filter(
-                document_page=self.test_document.pages.first()
+                document_version_page=self.test_document.pages.first().content_object
             ).exists()
         )
 
-    def test_document_page_content_view_no_permissions(self):
-        self.test_document.submit_for_ocr()
-
-        response = self._request_document_page_content_view()
-        self.assertEqual(response.status_code, 404)
-
-    def test_document_page_content_view_with_access(self):
-        self.test_document.submit_for_ocr()
-        self.grant_access(
-            obj=self.test_document, permission=permission_ocr_content_view
-        )
-
-        response = self._request_document_page_content_view()
-        self.assertContains(
-            response=response, text=TEST_DOCUMENT_CONTENT, status_code=200
-        )
-
     def test_document_submit_view_no_permission(self):
         response = self._request_document_submit_view()
         self.assertEqual(response.status_code, 404)
@@ -188,6 +171,23 @@ class OCRViewsTestCase(OCRViewTestMixin, GenericDocumentViewTestCase):
             ),
         )
 
+    def test_document_version_page_content_view_no_permissions(self):
+        self.test_document.submit_for_ocr()
+
+        response = self._request_document_version_page_content_view()
+        self.assertEqual(response.status_code, 404)
+
+    def test_document_version_page_content_view_with_access(self):
+        self.test_document.submit_for_ocr()
+        self.grant_access(
+            obj=self.test_document, permission=permission_ocr_content_view
+        )
+
+        response = self._request_document_version_page_content_view()
+        self.assertContains(
+            response=response, text=TEST_DOCUMENT_CONTENT, status_code=200
+        )
+
 
 class DocumentTypeOCRViewTestMixin(object):
     def _request_document_type_ocr_settings_view(self):

mayan/apps/ocr/urls.py

@@ -8,9 +8,10 @@ from .api_views import (
 )
 from .views import (
     DocumentOCRContentDeleteView, DocumentOCRContentView,
-    DocumentOCRDownloadView,
+    DocumentOCRDownloadView, DocumentOCRErrorsListView,
-    DocumentOCRErrorsListView, DocumentPageOCRContentView, DocumentSubmitView,
+    DocumentPageOCRContentView, DocumentSubmitView,
-    DocumentTypeSettingsEditView, DocumentTypeSubmitView, EntryListView
+    DocumentTypeSettingsEditView, DocumentTypeSubmitView,
+    DocumentVersionPageOCRContentView, EntryListView
 )
 
 urlpatterns = [
@@ -50,6 +51,11 @@ urlpatterns = [
         view=DocumentPageOCRContentView.as_view(),
         name='document_page_ocr_content'
     ),
+    url(
+        regex=r'^documents/versions/pages/(?P<pk>\d+)/content/$',
+        view=DocumentVersionPageOCRContentView.as_view(),
+        name='document_version_page_ocr_content'
+    ),
     url(
         regex=r'^document_types/submit/$',
         view=DocumentTypeSubmitView.as_view(), name='document_type_submit'

mayan/apps/ocr/utils.py

@@ -4,15 +4,29 @@ from django.apps import apps
 from django.utils.encoding import force_text
 
 
-def get_document_ocr_content(document):
+def get_document_version_ocr_content(document_version):
-    DocumentPageOCRContent = apps.get_model(
+    DocumentVersionPageOCRContent = apps.get_model(
-        app_label='ocr', model_name='DocumentPageOCRContent'
+        app_label='ocr', model_name='DocumentVersionPageOCRContent'
     )
 
-    for page in document.pages.all():
+    for document_version_page in document_version.pages.all():
         try:
-            page_content = page.ocr_content.content
+            page_content = document_version_page.ocr_content.content
-        except DocumentPageOCRContent.DoesNotExist:
+        except DocumentVersionPageOCRContent.DoesNotExist:
+            pass
+        else:
+            yield force_text(page_content)
+
+
+def get_document_ocr_content(document):
+    DocumentVersionPageOCRContent = apps.get_model(
+        app_label='ocr', model_name='DocumentVersionPageOCRContent'
+    )
+
+    for document_page in document.pages.all():
+        try:
+            page_content = document_page.content_object.ocr_content.content
+        except (AttributeError, DocumentVersionPageOCRContent.DoesNotExist):
             pass
         else:
             yield force_text(page_content)

mayan/apps/ocr/views.py

@@ -12,10 +12,15 @@ from mayan.apps.common.generics import (
 )
 from mayan.apps.common.mixins import ExternalObjectMixin
 from mayan.apps.documents.forms import DocumentTypeFilteredSelectForm
-from mayan.apps.documents.models import Document, DocumentPage, DocumentType
+from mayan.apps.documents.models import (
+    Document, DocumentPage, DocumentType, DocumentVersionPage
+)
 
-from .forms import DocumentPageOCRContentForm, DocumentOCRContentForm
+from .forms import (
-from .models import DocumentPageOCRContent, DocumentVersionOCRError
+    DocumentPageOCRContentForm, DocumentOCRContentForm,
+    DocumentVersionPageOCRContentForm
+)
+from .models import DocumentVersionPageOCRContent, DocumentVersionOCRError
 from .permissions import (
     permission_ocr_content_view, permission_ocr_document,
     permission_document_type_ocr_setup
@@ -46,7 +51,7 @@ class DocumentOCRContentDeleteView(MultipleObjectConfirmActionView):
         return result
 
     def object_action(self, form, instance):
-        DocumentPageOCRContent.objects.delete_content_for(
+        DocumentVersionPageOCRContent.objects.delete_content_for(
             document=instance, user=self.request.user
         )
 
@@ -94,6 +99,30 @@ class DocumentPageOCRContentView(SingleObjectDetailView):
         }
 
 
+class DocumentVersionPageOCRContentView(SingleObjectDetailView):
+    form_class = DocumentVersionPageOCRContentForm
+    model = DocumentVersionPage
+    object_permission = permission_ocr_content_view
+
+    def dispatch(self, request, *args, **kwargs):
+        result = super(DocumentVersionPageOCRContentView, self).dispatch(
+            request, *args, **kwargs
+        )
+        self.get_object().document.add_as_recent_document_for_user(
+            user=request.user
+        )
+        return result
+
+    def get_extra_context(self):
+        return {
+            'hide_labels': True,
+            'object': self.get_object(),
+            'title': _(
+                'OCR result for document version page: %s'
+            ) % self.get_object(),
+        }
+
+
 class DocumentSubmitView(MultipleObjectConfirmActionView):
     model = Document
     object_permission = permission_ocr_document

mayan/apps/sources/apps.py

@@ -21,9 +21,10 @@ from .handlers import (
     handler_create_default_document_source, handler_initialize_periodic_tasks
 )
 from .links import (
-    link_document_create_multiple, link_setup_sources,
+    link_document_create_multiple, link_document_pages_append,
-    link_setup_source_check_now, link_setup_source_create_imap_email,
+    link_setup_sources, link_setup_source_check_now,
-    link_setup_source_create_pop3_email, link_setup_source_create_sane_scanner,
+    link_setup_source_create_imap_email, link_setup_source_create_pop3_email,
+    link_setup_source_create_sane_scanner,
     link_setup_source_create_watch_folder, link_setup_source_create_webform,
     link_setup_source_create_staging_folder, link_setup_source_delete,
     link_setup_source_edit, link_setup_source_logs, link_staging_file_delete,
@@ -145,10 +146,16 @@ class SourcesApp(MayanAppConfig):
         menu_secondary.bind_links(
             links=(link_document_version_upload,),
             sources=(
-                'documents:document_version_list', 'documents:upload_version',
+                'documents:document_version_list', 'sources:upload_version',
                 'documents:document_version_revert'
             )
         )
+        menu_secondary.bind_links(
+            links=(link_document_pages_append,),
+            sources=(
+                'documents:document_pages', 'sources:document_pages_append'
+            )
+        )
 
         post_upgrade.connect(
             receiver=handler_initialize_periodic_tasks,

mayan/apps/sources/forms.py

@@ -23,14 +23,25 @@ class NewDocumentForm(DocumentForm):
 
 
 class NewVersionForm(forms.Form):
-    def __init__(self, *args, **kwargs):
+    comment = forms.CharField(
-        super(NewVersionForm, self).__init__(*args, **kwargs)
+        help_text=_('An optional comment to explain the upload.'),
+        label=_('Comment'), required=False,
+        widget=forms.widgets.Textarea(attrs={'rows': 4}),
+    )
 
-        self.fields['comment'] = forms.CharField(
+    append_pages = forms.BooleanField(
-            label=_('Comment'),
+        help_text=_(
-            required=False,
+            'If selected, the pages of the file uploaded will be appended '
-            widget=forms.widgets.Textarea(attrs={'rows': 4}),
+            'to the existing document pages. Otherwise the pages of the '
-        )
+            'upload will replace the existing pages of the document.'
+        ), label=_('Append pages?'), required=False,
+    )
+
+    def __init__(self, *args, **kwargs):
+        hide_append_pages = kwargs.pop('hide_append_pages', False)
+        super(NewVersionForm, self).__init__(*args, **kwargs)
+        if hide_append_pages:
+            self.fields['append_pages'].widget = forms.widgets.HiddenInput()
 
 
 class UploadBaseForm(forms.Form):

mayan/apps/sources/icons.py

@@ -9,6 +9,10 @@ icon_document_version_upload = Icon(
     driver_name='fontawesome', symbol='upload'
 )
 icon_log = Icon(driver_name='fontawesome', symbol='exclamation-triangle')
+icon_document_pages_append = Icon(
+    driver_name='fontawesome-dual', primary_symbol='copy',
+    secondary_symbol='plus'
+)
 icon_setup_sources = Icon(driver_name='fontawesome', symbol='upload')
 icon_setup_source_check_now = Icon(driver_name='fontawesome', symbol='check')
 icon_setup_source_delete = Icon(driver_name='fontawesome', symbol='times')

mayan/apps/sources/links.py

@@ -113,6 +113,13 @@ link_staging_file_delete = Link(
     permissions=(permission_document_new_version, permission_document_create),
     tags='dangerous', text=_('Delete'), view='sources:staging_file_delete',
 )
+link_document_pages_append = Link(
+    args='resolved_object.pk',
+    icon_class_path='mayan.apps.sources.icons.icon_document_pages_append',
+    permissions=(permission_document_new_version,),
+    text=_('Append pages'),
+    view='sources:document_pages_append'
+)
 link_document_version_upload = Link(
     args='resolved_object.pk', condition=document_new_version_not_blocked,
     icon_class_path='mayan.apps.sources.icons.icon_document_version_upload',

mayan/apps/sources/urls.py

@@ -7,9 +7,10 @@ from .api_views import (
     APIStagingSourceListView, APIStagingSourceView
 )
 from .views import (
-    SetupSourceCheckView, SetupSourceCreateView, SetupSourceDeleteView,
+    DocumentPagesAppendView, SetupSourceCheckView, SetupSourceCreateView,
-    SetupSourceEditView, SetupSourceListView, SourceLogListView,
+    SetupSourceDeleteView, SetupSourceEditView, SetupSourceListView,
-    StagingFileDeleteView, UploadInteractiveVersionView, UploadInteractiveView
+    SourceLogListView, StagingFileDeleteView, UploadInteractiveVersionView,
+    UploadInteractiveView
 )
 from .wizards import DocumentCreateWizard
 
@@ -41,6 +42,14 @@ urlpatterns = [
         regex=r'^documents/(?P<document_pk>\d+)/versions/upload/interactive/$',
         view=UploadInteractiveVersionView.as_view(), name='upload_version'
     ),
+    url(
+        regex=r'^documents/(?P<document_pk>\d+)/pages/append/interactive/(?P<source_id>\d+)/$',
+        view=DocumentPagesAppendView.as_view(), name='document_pages_append'
+    ),
+    url(
+        regex=r'^documents/(?P<document_pk>\d+)/pages/append/interactive/$',
+        view=DocumentPagesAppendView.as_view(), name='document_pages_append'
+    ),
 
     # Setup views
 

mayan/apps/sources/views.py

@@ -362,7 +362,6 @@ class UploadInteractiveView(UploadBaseView):
 
 class UploadInteractiveVersionView(UploadBaseView):
     def dispatch(self, request, *args, **kwargs):
-
         self.subtemplates_list = []
 
         self.document = get_object_or_404(
@@ -417,12 +416,17 @@ class UploadInteractiveVersionView(UploadBaseView):
             else:
                 user_id = None
 
-            task_upload_new_version.apply_async(kwargs=dict(
+            task_upload_new_version.apply_async(
-                shared_uploaded_file_id=shared_uploaded_file.pk,
+                kwargs=dict(
-                document_id=self.document.pk,
+                    append_pages=forms['document_form'].cleaned_data.get(
-                user_id=user_id,
+                        'append_pages', False
-                comment=forms['document_form'].cleaned_data.get('comment')
+                    ),
-            ))
+                    shared_uploaded_file_id=shared_uploaded_file.pk,
+                    document_id=self.document.pk,
+                    user_id=user_id,
+                    comment=forms['document_form'].cleaned_data.get('comment')
+                )
+            )
 
             messages.success(
                 message=_(
@@ -448,13 +452,6 @@ class UploadInteractiveVersionView(UploadBaseView):
             files=kwargs.get('files', None),
         )
 
-    def create_document_form_form(self, **kwargs):
-        return self.get_form_classes()['document_form'](
-            prefix=kwargs['prefix'],
-            data=kwargs.get('data', None),
-            files=kwargs.get('files', None),
-        )
-
     def get_form_classes(self):
         return {
             'document_form': NewVersionForm,
@@ -467,8 +464,33 @@ class UploadInteractiveVersionView(UploadBaseView):
         ).get_context_data(**kwargs)
         context['object'] = self.document
         context['title'] = _(
-            'Upload a new version from source: %s'
+            'Upload a new version for document "%(document)s" from source: %(source)s'
-        ) % self.source.label
+        ) % {'document': self.document, 'source': self.source.label}
+
+        context['submit_label'] = _('Submit')
+
+        return context
+
+
+class DocumentPagesAppendView(UploadInteractiveVersionView):
+    def get_document_form_initial(self):
+        return {
+            'append_pages': True,
+        }
+
+    def get_form_extra_kwargs(self, form_name):
+        if form_name == 'document_form':
+            return {
+                'hide_append_pages': True
+            }
+
+    def get_context_data(self, **kwargs):
+        context = super(
+            DocumentPagesAppendView, self
+        ).get_context_data(**kwargs)
+        context['title'] = _(
+            'Append pages to document "%(document)s" from source: %(source)s'
+        ) % {'document': self.document, 'source': self.source.label}
 
         return context
 

mayan/apps/tags/apps.py

@@ -13,7 +13,9 @@ from mayan.apps.common.menus import (
     menu_facet, menu_list_facet, menu_main, menu_multi_item, menu_object,
     menu_secondary
 )
-from mayan.apps.documents.search import document_page_search, document_search
+from mayan.apps.documents.search import (
+    document_page_search, document_search, document_version_page_search
+)
 from mayan.apps.events.classes import ModelEventType
 from mayan.apps.events.links import (
     link_events_for_object, link_object_event_types_user_subcriptions_list,
@@ -62,7 +64,7 @@ class TagsApp(MayanAppConfig):
         )
 
         DocumentPageResult = apps.get_model(
-            app_label='documents', model_name='DocumentPageResult'
+            app_label='documents', model_name='DocumentVersionPageResult'
         )
 
         DocumentTag = self.get_model(model_name='DocumentTag')
@@ -133,9 +135,12 @@ class TagsApp(MayanAppConfig):
         )
 
         document_page_search.add_model_field(
-            field='document_version__document__tags__label', label=_('Tags')
+            field='document__tags__label', label=_('Tags')
         )
         document_search.add_model_field(field='tags__label', label=_('Tags'))
+        document_version_page_search.add_model_field(
+            field='document_version__document__tags__label', label=_('Tags')
+        )
 
         menu_facet.bind_links(
             links=(link_document_tag_list,), sources=(Document,)

requirements/base.txt

@@ -1,9 +1,7 @@
 Pillow==6.0.0
 PyPDF2==1.26.0
 PyYAML==5.1.1
-celery==4.3.0
 django-activity-stream==0.7.0
-django-celery-beat==1.5.0
 django-colorful==1.3
 django-cors-headers==2.5.2
 django-downloadview==1.9

setup.py

@@ -60,9 +60,7 @@ django==1.11.24
 Pillow==6.0.0
 PyPDF2==1.26.0
 PyYAML==5.1.1
-celery==4.3.0
 django-activity-stream==0.7.0
-django-celery-beat==1.5.0
 django-colorful==1.3
 django-cors-headers==2.5.2
 django-downloadview==1.9